wip: mermaid

* fix: markdown rehype highlight accidental removal

* fix: USE_REDIS condition check

* fix: markdown list counter

.env.example

@@ -64,6 +64,9 @@ PROXY=
 
 # ANYSCALE_API_KEY=
 # APIPIE_API_KEY=
+# COHERE_API_KEY=
+# DEEPSEEK_API_KEY=
+# DATABRICKS_API_KEY=
 # FIREWORKS_API_KEY=
 # GROQ_API_KEY=
 # HUGGINGFACE_TOKEN=
@@ -72,20 +75,20 @@ PROXY=
 # PERPLEXITY_API_KEY=
 # SHUTTLEAI_API_KEY=
 # TOGETHERAI_API_KEY=
+# UNIFY_API_KEY=
 
 #============#
 # Anthropic  #
 #============#
 
 ANTHROPIC_API_KEY=user_provided
-# ANTHROPIC_MODELS=claude-3-opus-20240229,claude-3-sonnet-20240229,claude-3-haiku-20240307,claude-2.1,claude-2,claude-1.2,claude-1,claude-1-100k,claude-instant-1,claude-instant-1-100k
+# ANTHROPIC_MODELS=claude-3-5-sonnet-20240620,claude-3-opus-20240229,claude-3-sonnet-20240229,claude-3-haiku-20240307,claude-2.1,claude-2,claude-1.2,claude-1,claude-1-100k,claude-instant-1,claude-instant-1-100k
 # ANTHROPIC_REVERSE_PROXY=
 
 #============#
 # Azure      #
 #============#
 
-
 # Note: these variables are DEPRECATED
 # Use the `librechat.yaml` configuration for `azureOpenAI` instead
 # You may also continue to use them if you opt out of using the `librechat.yaml` configuration
@@ -115,32 +118,38 @@ BINGAI_TOKEN=user_provided
 GOOGLE_KEY=user_provided
 # GOOGLE_REVERSE_PROXY=
 
-# Gemini API
+# Gemini API (AI Studio)
 # GOOGLE_MODELS=gemini-1.5-flash-latest,gemini-1.0-pro,gemini-1.0-pro-001,gemini-1.0-pro-latest,gemini-1.0-pro-vision-latest,gemini-1.5-pro-latest,gemini-pro,gemini-pro-vision
 
 # Vertex AI
-# GOOGLE_MODELS=gemini-1.5-flash-preview-0514,gemini-1.5-pro-preview-0409,gemini-1.0-pro-vision-001,gemini-pro,gemini-pro-vision,chat-bison,chat-bison-32k,codechat-bison,codechat-bison-32k,text-bison,text-bison-32k,text-unicorn,code-gecko,code-bison,code-bison-32k
+# GOOGLE_MODELS=gemini-1.5-flash-preview-0514,gemini-1.5-pro-preview-0514,gemini-1.0-pro-vision-001,gemini-1.0-pro-002,gemini-1.0-pro-001,gemini-pro-vision,gemini-1.0-pro
 
-# Google Gemini Safety Settings
-# NOTE (Vertex AI): You do not have access to the BLOCK_NONE setting by default.
-# To use this restricted HarmBlockThreshold setting, you will need to either:
+# GOOGLE_TITLE_MODEL=gemini-pro
+
+# Google Safety Settings
+# NOTE: These settings apply to both Vertex AI and Gemini API (AI Studio)
 #
-# (a) Get access through an allowlist via your Google account team
-# (b) Switch your account type to monthly invoiced billing following this instruction:
-#     https://cloud.google.com/billing/docs/how-to/invoiced-billing
+# For Vertex AI:
+# To use the BLOCK_NONE setting, you need either:
+# (a) Access through an allowlist via your Google account team, or
+# (b) Switch to monthly invoiced billing: https://cloud.google.com/billing/docs/how-to/invoiced-billing
+#
+# For Gemini API (AI Studio):
+# BLOCK_NONE is available by default, no special account requirements.
+#
+# Available options: BLOCK_NONE, BLOCK_ONLY_HIGH, BLOCK_MEDIUM_AND_ABOVE, BLOCK_LOW_AND_ABOVE
 #
 # GOOGLE_SAFETY_SEXUALLY_EXPLICIT=BLOCK_ONLY_HIGH
 # GOOGLE_SAFETY_HATE_SPEECH=BLOCK_ONLY_HIGH
 # GOOGLE_SAFETY_HARASSMENT=BLOCK_ONLY_HIGH
 # GOOGLE_SAFETY_DANGEROUS_CONTENT=BLOCK_ONLY_HIGH
 
-
 #============#
 # OpenAI     #
 #============#
 
 OPENAI_API_KEY=user_provided
-# OPENAI_MODELS=gpt-4o,gpt-3.5-turbo-0125,gpt-3.5-turbo-0301,gpt-3.5-turbo,gpt-4,gpt-4-0613,gpt-4-vision-preview,gpt-3.5-turbo-0613,gpt-3.5-turbo-16k-0613,gpt-4-0125-preview,gpt-4-turbo-preview,gpt-4-1106-preview,gpt-3.5-turbo-1106,gpt-3.5-turbo-instruct,gpt-3.5-turbo-instruct-0914,gpt-3.5-turbo-16k
+# OPENAI_MODELS=gpt-4o,chatgpt-4o-latest,gpt-4o-mini,gpt-3.5-turbo-0125,gpt-3.5-turbo-0301,gpt-3.5-turbo,gpt-4,gpt-4-0613,gpt-4-vision-preview,gpt-3.5-turbo-0613,gpt-3.5-turbo-16k-0613,gpt-4-0125-preview,gpt-4-turbo-preview,gpt-4-1106-preview,gpt-3.5-turbo-1106,gpt-3.5-turbo-instruct,gpt-3.5-turbo-instruct-0914,gpt-3.5-turbo-16k
 
 DEBUG_OPENAI=false
 
@@ -162,7 +171,17 @@ DEBUG_OPENAI=false
 
 ASSISTANTS_API_KEY=user_provided
 # ASSISTANTS_BASE_URL=
-# ASSISTANTS_MODELS=gpt-4o,gpt-3.5-turbo-0125,gpt-3.5-turbo-16k-0613,gpt-3.5-turbo-16k,gpt-3.5-turbo,gpt-4,gpt-4-0314,gpt-4-32k-0314,gpt-4-0613,gpt-3.5-turbo-0613,gpt-3.5-turbo-1106,gpt-4-0125-preview,gpt-4-turbo-preview,gpt-4-1106-preview
+# ASSISTANTS_MODELS=gpt-4o,gpt-4o-mini,gpt-3.5-turbo-0125,gpt-3.5-turbo-16k-0613,gpt-3.5-turbo-16k,gpt-3.5-turbo,gpt-4,gpt-4-0314,gpt-4-32k-0314,gpt-4-0613,gpt-3.5-turbo-0613,gpt-3.5-turbo-1106,gpt-4-0125-preview,gpt-4-turbo-preview,gpt-4-1106-preview
+
+#==========================#
+#   Azure Assistants API   #
+#==========================#
+
+# Note: You should map your credentials with custom variables according to your Azure OpenAI Configuration
+# The models for Azure Assistants are also determined by your Azure OpenAI configuration.
+
+# More info, including how to enable use of Assistants with Azure here:
+# https://www.librechat.ai/docs/configuration/librechat_yaml/ai_endpoints/azure#using-assistants-with-azure
 
 #============#
 # OpenRouter #
@@ -174,7 +193,7 @@ ASSISTANTS_API_KEY=user_provided
 # Plugins    #
 #============#
 
-# PLUGIN_MODELS=gpt-4o,gpt-4,gpt-4-turbo-preview,gpt-4-0125-preview,gpt-4-1106-preview,gpt-4-0613,gpt-3.5-turbo,gpt-3.5-turbo-0125,gpt-3.5-turbo-1106,gpt-3.5-turbo-0613
+# PLUGIN_MODELS=gpt-4o,gpt-4o-mini,gpt-4,gpt-4-turbo-preview,gpt-4-0125-preview,gpt-4-1106-preview,gpt-4-0613,gpt-3.5-turbo,gpt-3.5-turbo-0125,gpt-3.5-turbo-1106,gpt-3.5-turbo-0613
 
 DEBUG_PLUGINS=true
 
@@ -247,6 +266,23 @@ MEILI_NO_ANALYTICS=true
 MEILI_HOST=http://0.0.0.0:7700
 MEILI_MASTER_KEY=DrhYf7zENyR6AlUCKmnz0eYASOQdl6zxH7s7MKFSfFCt
 
+#==================================================#
+#          Speech to Text & Text to Speech         #
+#==================================================#
+
+STT_API_KEY=
+TTS_API_KEY=
+
+#==================================================#
+#                        RAG                       #
+#==================================================#
+# More info: https://www.librechat.ai/docs/configuration/rag_api
+
+# RAG_OPENAI_BASEURL=
+# RAG_OPENAI_API_KEY=
+# EMBEDDINGS_PROVIDER=openai
+# EMBEDDINGS_MODEL=text-embedding-3-small
+
 #===================================================#
 #                    User System                    #
 #===================================================#
@@ -301,6 +337,9 @@ ALLOW_EMAIL_LOGIN=true
 ALLOW_REGISTRATION=true
 ALLOW_SOCIAL_LOGIN=false
 ALLOW_SOCIAL_REGISTRATION=false
+ALLOW_PASSWORD_RESET=false
+# ALLOW_ACCOUNT_DELETION=true # note: enabled by default if omitted/commented out
+ALLOW_UNVERIFIED_EMAIL_LOGIN=true
 
 SESSION_EXPIRY=1000 * 60 * 15
 REFRESH_TOKEN_EXPIRY=(1000 * 60 * 60 * 24) * 7
@@ -342,6 +381,19 @@ OPENID_REQUIRED_ROLE_PARAMETER_PATH=
 OPENID_BUTTON_LABEL=
 OPENID_IMAGE_URL=
 
+# LDAP
+LDAP_URL=
+LDAP_BIND_DN=
+LDAP_BIND_CREDENTIALS=
+LDAP_USER_SEARCH_BASE=
+LDAP_SEARCH_FILTER=mail={{username}}
+LDAP_CA_CERT_PATH=
+# LDAP_TLS_REJECT_UNAUTHORIZED=
+# LDAP_LOGIN_USES_USERNAME=true
+# LDAP_ID=
+# LDAP_USERNAME=
+# LDAP_FULL_NAME=
+
 #========================#
 # Email Password Reset   #
 #========================#
@@ -368,6 +420,25 @@ FIREBASE_STORAGE_BUCKET=
 FIREBASE_MESSAGING_SENDER_ID=
 FIREBASE_APP_ID=
 
+#========================#
+# Shared Links           #
+#========================#
+
+ALLOW_SHARED_LINKS=true
+ALLOW_SHARED_LINKS_PUBLIC=true
+
+#==============================#
+# Static File Cache Control    #
+#==============================#
+
+# Leave commented out to use defaults: 1 day (86400 seconds) for s-maxage and 2 days (172800 seconds) for max-age
+# NODE_ENV must be set to production for these to take effect
+# STATIC_CACHE_MAX_AGE=172800
+# STATIC_CACHE_S_MAX_AGE=86400
+
+# If you have another service in front of your LibreChat doing compression, disable express based compression here
+# DISABLE_COMPRESSION=true
+
 #===================================================#
 #                        UI                         #
 #===================================================#
@@ -378,6 +449,9 @@ HELP_AND_FAQ_URL=https://librechat.ai
 
 # SHOW_BIRTHDAY_ICON=true
 
+# Google tag manager id
+#ANALYTICS_GTM_ID=user provided google tag manager id
+
 #==================================================#
 #                      Others                      #
 #==================================================#

.eslintrc.js

@@ -12,6 +12,7 @@ module.exports = {
     'plugin:react-hooks/recommended',
     'plugin:jest/recommended',
     'prettier',
+    'plugin:jsx-a11y/recommended',
   ],
   ignorePatterns: [
     'client/dist/**/*',
@@ -32,7 +33,7 @@ module.exports = {
       jsx: true,
     },
   },
-  plugins: ['react', 'react-hooks', '@typescript-eslint', 'import'],
+  plugins: ['react', 'react-hooks', '@typescript-eslint', 'import', 'jsx-a11y'],
   rules: {
     'react/react-in-jsx-scope': 'off',
     '@typescript-eslint/ban-ts-comment': ['error', { 'ts-ignore': 'allow' }],
@@ -65,6 +66,7 @@ module.exports = {
     'no-restricted-syntax': 'off',
     'react/prop-types': ['off'],
     'react/display-name': ['off'],
+    'no-nested-ternary': 'error',
     'no-unused-vars': ['error', { varsIgnorePattern: '^_' }],
     quotes: ['error', 'single'],
   },
@@ -118,6 +120,8 @@ module.exports = {
       ],
       rules: {
         '@typescript-eslint/no-explicit-any': 'error',
+        '@typescript-eslint/no-unnecessary-condition': 'warn',
+        '@typescript-eslint/strict-boolean-expressions': 'warn',
       },
     },
     {

.github/CONTRIBUTING.md

@@ -126,6 +126,18 @@ Apply the following naming conventions to branches, labels, and other Git-relate
       
    - **Current Stance**: At present, this backend transition is of lower priority and might not be pursued.
 
+## 7. Module Import Conventions
+
+- `npm` packages first, 
+     - from shortest line (top) to longest (bottom)
+
+- Followed by typescript types (pertains to data-provider and client workspaces)
+     - longest line (top) to shortest (bottom)
+     - types from package come first
+
+- Lastly, local imports
+     - longest line (top) to shortest (bottom)
+     - imports with alias `~` treated the same as relative import with respect to line length
 
 ---
 

.github/workflows/a11y.yml

@@ -0,0 +1,26 @@
+name: Lint for accessibility issues
+
+on:
+  pull_request:
+    paths:
+      - 'client/src/**'
+  workflow_dispatch:
+    inputs:
+      run_workflow:
+        description: 'Set to true to run this workflow'
+        required: true
+        default: 'false'
+
+jobs:
+  axe-linter:
+    runs-on: ubuntu-latest
+    if: >
+      (github.event_name == 'pull_request' && github.event.pull_request.head.repo.full_name == 'danny-avila/LibreChat') ||
+      (github.event_name == 'workflow_dispatch' && github.event.inputs.run_workflow == 'true')
+    
+    steps:
+      - uses: actions/checkout@v4
+      - uses: dequelabs/axe-linter-action@v1
+        with:
+          api_key: ${{ secrets.AXE_LINTER_API_KEY }}
+          github_token: ${{ secrets.GITHUB_TOKEN }}

.github/workflows/deploy-dev.yml

@@ -0,0 +1,41 @@
+name: Update Test Server
+
+on:
+  workflow_run:
+    workflows: ["Docker Dev Images Build"]
+    types:
+      - completed
+  workflow_dispatch:
+
+jobs:
+  deploy:
+    runs-on: ubuntu-latest
+    if: |
+      github.repository == 'danny-avila/LibreChat' &&
+      (github.event_name == 'workflow_dispatch' || github.event.workflow_run.conclusion == 'success')
+    steps:
+    - name: Checkout repository
+      uses: actions/checkout@v4
+
+    - name: Install SSH Key
+      uses: shimataro/ssh-key-action@v2
+      with:
+        key: ${{ secrets.DO_SSH_PRIVATE_KEY }}
+        known_hosts: ${{ secrets.DO_KNOWN_HOSTS }}
+
+    - name: Run update script on DigitalOcean Droplet
+      env:
+        DO_HOST: ${{ secrets.DO_HOST }}
+        DO_USER: ${{ secrets.DO_USER }}
+      run: |
+        ssh -o StrictHostKeyChecking=no ${DO_USER}@${DO_HOST} << EOF
+        sudo -i -u danny bash << EEOF
+        cd ~/LibreChat && \
+        git fetch origin main && \
+        npm run update:deployed && \
+        git checkout do-deploy && \
+        git rebase main && \
+        npm run start:deployed && \
+        echo "Update completed. Application should be running now."
+        EEOF
+        EOF

.github/workflows/helmcharts.yml

@@ -0,0 +1,35 @@
+name: Build Helm Charts on Tag
+
+# The workflow is triggered when a tag is pushed
+on:
+  push:
+    tags:
+      - "*"
+
+jobs:
+  release:
+    permissions:
+      contents: write
+    runs-on: ubuntu-latest
+    steps:
+      - name: Checkout
+        uses: actions/checkout@v4
+        with:
+          fetch-depth: 0
+
+      - name: Configure Git
+        run: |
+          git config user.name "$GITHUB_ACTOR"
+          git config user.email "$GITHUB_ACTOR@users.noreply.github.com"
+
+      - name: Install Helm
+        uses: azure/setup-helm@v4
+        env:
+          GITHUB_TOKEN: "${{ secrets.GITHUB_TOKEN }}"       
+
+      - name: Run chart-releaser
+        uses: helm/chart-releaser-action@v1.6.0
+        with:
+          charts_dir: helmchart
+        env:
+          CR_TOKEN: "${{ secrets.GITHUB_TOKEN }}"

.gitignore

@@ -11,6 +11,7 @@ logs
 pids
 *.pid
 *.seed
+.git
 
 # Directory for instrumented libs generated by jscoverage/JSCover
 lib-cov
@@ -45,6 +46,7 @@ api/node_modules/
 client/node_modules/
 bower_components/
 *.d.ts
+!vite-env.d.ts
 
 # Floobits
 .floo

.vscode/launch.json

@@ -0,0 +1,16 @@
+{
+  "version": "0.2.0",
+  "configurations": [
+    {
+      "type": "node",
+      "request": "launch",
+      "name": "Launch LibreChat (debug)",
+      "skipFiles": ["<node_internals>/**"],
+      "program": "${workspaceFolder}/api/server/index.js",
+      "env": {
+        "NODE_ENV": "production"
+      },
+      "console": "integratedTerminal"
+    }
+  ]
+}

Dockerfile

@@ -1,4 +1,4 @@
-# v0.7.2
+# v0.7.5-rc1
 
 # Base node image
 FROM node:20-alpine AS node

Dockerfile.multi

@@ -1,43 +1,44 @@
-# v0.7.2
+# Dockerfile.multi
+# v0.7.5-rc1
 
-# Build API, Client and Data Provider
+# Base for all builds
 FROM node:20-alpine AS base
+WORKDIR /app
+RUN apk --no-cache add curl
+RUN npm config set fetch-retry-maxtimeout 600000 && \
+    npm config set fetch-retries 5 && \
+    npm config set fetch-retry-mintimeout 15000
+COPY package*.json ./
+COPY packages/data-provider/package*.json ./packages/data-provider/
+COPY client/package*.json ./client/
+COPY api/package*.json ./api/
+RUN npm ci
 
 # Build data-provider
 FROM base AS data-provider-build
 WORKDIR /app/packages/data-provider
-COPY ./packages/data-provider ./
-RUN npm install; npm cache clean --force
+COPY packages/data-provider ./
 RUN npm run build
 RUN npm prune --production
 
-# React client build
+# Client build
 FROM base AS client-build
 WORKDIR /app/client
-COPY ./client/package*.json ./
-# Copy data-provider to client's node_modules
-COPY --from=data-provider-build /app/packages/data-provider/ /app/client/node_modules/librechat-data-provider/
-RUN npm install; npm cache clean --force
-COPY ./client/ ./
+COPY client ./
+COPY --from=data-provider-build /app/packages/data-provider/dist /app/packages/data-provider/dist
 ENV NODE_OPTIONS="--max-old-space-size=2048"
 RUN npm run build
+RUN npm prune --production
 
-# Node API setup
+# API setup (including client dist)
 FROM base AS api-build
+WORKDIR /app
+COPY api ./api
+COPY config ./config
+COPY --from=data-provider-build /app/packages/data-provider/dist ./packages/data-provider/dist
+COPY --from=client-build /app/client/dist ./client/dist
 WORKDIR /app/api
-COPY api/package*.json ./
-COPY api/ ./
-# Copy helper scripts
-COPY config/ ./
-# Copy data-provider to API's node_modules
-COPY --from=data-provider-build /app/packages/data-provider/ /app/api/node_modules/librechat-data-provider/
-RUN npm install --include prod; npm cache clean --force
-COPY --from=client-build /app/client/dist /app/client/dist
+RUN npm prune --production
 EXPOSE 3080
 ENV HOST=0.0.0.0
 CMD ["node", "server/index.js"]
-
-# Nginx setup
-FROM nginx:1.21.1-alpine AS prod-stage
-COPY ./client/nginx.conf /etc/nginx/conf.d/default.conf
-CMD ["nginx", "-g", "daemon off;"]

README.md

@@ -27,7 +27,7 @@
 </p>
 
 <p align="center">
-<a href="https://railway.app/template/b5k2mn?referralCode=myKrVZ">
+<a href="https://railway.app/template/b5k2mn?referralCode=HI9hWz">
   <img src="https://railway.app/button.svg" alt="Deploy on Railway" height="30">
 </a>
 <a href="https://zeabur.com/templates/0X2ZY8">
@@ -50,7 +50,7 @@
 - 🔄 Edit, Resubmit, and Continue Messages with Conversation branching
 - 🌿 Fork Messages & Conversations for Advanced Context control
 - 💬 Multimodal Chat:
-    - Upload and analyze images with Claude 3, GPT-4 (including `gpt-4o`), and Gemini Vision 📸
+    - Upload and analyze images with Claude 3, GPT-4 (including `gpt-4o` and `gpt-4o-mini`), and Gemini Vision 📸
     - Chat with Files using Custom Endpoints, OpenAI, Azure, Anthropic, & Google. 🗃️
     - Advanced Agents with Files, Code Interpreter, Tools, and API Actions 🔦
       - Available through the [OpenAI Assistants API](https://platform.openai.com/docs/assistants/overview) 🌤️
@@ -58,9 +58,13 @@
 - 🌎 Multilingual UI:
   - English, 中文, Deutsch, Español, Français, Italiano, Polski, Português Brasileiro,
   - Русский, 日本語, Svenska, 한국어, Tiếng Việt, 繁體中文, العربية, Türkçe, Nederlands, עברית
-- 🎨 Customizable Dropdown & Interface: Adapts to both power users and newcomers.
+- 🎨 Customizable Dropdown & Interface: Adapts to both power users and newcomers
+- 📧 Verify your email to ensure secure access
+- 🗣️ Chat hands-free with Speech-to-Text and Text-to-Speech magic
+  - Automatically send and play Audio
+  - Supports OpenAI, Azure OpenAI, and Elevenlabs
 - 📥 Import Conversations from LibreChat, ChatGPT, Chatbot UI
-- 📤 Export conversations as screenshots, markdown, text, json.
+- 📤 Export conversations as screenshots, markdown, text, json
 - 🔍 Search all messages/conversations
 - 🔌 Plugins, including web access, image generation with DALL-E-3 and more
 - 👥 Multi-User, Secure Authentication with Moderation and Token spend tools
@@ -77,7 +81,7 @@ LibreChat brings together the future of assistant AIs with the revolutionary tec
 
 With LibreChat, you no longer need to opt for ChatGPT Plus and can instead use free or pay-per-call APIs. We welcome contributions, cloning, and forking to enhance the capabilities of this advanced chatbot platform.
 
-[![Watch the video](https://img.youtube.com/vi/YLVUW5UP9N0/maxresdefault.jpg)](https://www.youtube.com/watch?v=YLVUW5UP9N0)
+[![Watch the video](https://raw.githubusercontent.com/LibreChat-AI/librechat.ai/main/public/images/changelog/v0.7.4.png)](https://www.youtube.com/watch?v=cvosUxogdpI)
 Click on the thumbnail to open the video☝️
 
 ---

api/app/clients/AnthropicClient.js

@@ -1,20 +1,25 @@
 const Anthropic = require('@anthropic-ai/sdk');
+const { HttpsProxyAgent } = require('https-proxy-agent');
 const { encoding_for_model: encodingForModel, get_encoding: getEncoding } = require('tiktoken');
 const {
-  getResponseSender,
+  Constants,
   EModelEndpoint,
+  anthropicSettings,
+  getResponseSender,
   validateVisionModel,
 } = require('librechat-data-provider');
 const { encodeAndFormat } = require('~/server/services/Files/images/encode');
 const {
   truncateText,
   formatMessage,
+  addCacheControl,
   titleFunctionPrompt,
   parseParamFromPrompt,
   createContextHandlers,
 } = require('./prompts');
-const spendTokens = require('~/models/spendTokens');
-const { getModelMaxTokens } = require('~/utils');
+const { spendTokens, spendStructuredTokens } = require('~/models/spendTokens');
+const { getModelMaxTokens, matchModelName } = require('~/utils');
+const { sleep } = require('~/server/utils');
 const BaseClient = require('./BaseClient');
 const { logger } = require('~/config');
 
@@ -28,6 +33,9 @@ function delayBeforeRetry(attempts, baseDelay = 1000) {
   return new Promise((resolve) => setTimeout(resolve, baseDelay * attempts));
 }
 
+const tokenEventTypes = new Set(['message_start', 'message_delta']);
+const { legacy } = anthropicSettings;
+
 class AnthropicClient extends BaseClient {
   constructor(apiKey, options = {}) {
     super(apiKey, options);
@@ -38,6 +46,24 @@ class AnthropicClient extends BaseClient {
       ? options.contextStrategy.toLowerCase()
       : 'discard';
     this.setOptions(options);
+    /** @type {string | undefined} */
+    this.systemMessage;
+    /** @type {AnthropicMessageStartEvent| undefined} */
+    this.message_start;
+    /** @type {AnthropicMessageDeltaEvent| undefined} */
+    this.message_delta;
+    /** Whether the model is part of the Claude 3 Family
+     * @type {boolean} */
+    this.isClaude3;
+    /** Whether to use Messages API or Completions API
+     * @type {boolean} */
+    this.useMessages;
+    /** Whether or not the model is limited to the legacy amount of output tokens
+     * @type {boolean} */
+    this.isLegacyOutput;
+    /** Whether or not the model supports Prompt Caching
+     * @type {boolean} */
+    this.supportsCacheControl;
   }
 
   setOptions(options) {
@@ -57,18 +83,28 @@ class AnthropicClient extends BaseClient {
       this.options = options;
     }
 
-    const modelOptions = this.options.modelOptions || {};
-    this.modelOptions = {
-      ...modelOptions,
-      // set some good defaults (check for undefined in some cases because they may be 0)
-      model: modelOptions.model || 'claude-1',
-      temperature: typeof modelOptions.temperature === 'undefined' ? 1 : modelOptions.temperature, // 0 - 1, 1 is default
-      topP: typeof modelOptions.topP === 'undefined' ? 0.7 : modelOptions.topP, // 0 - 1, default: 0.7
-      topK: typeof modelOptions.topK === 'undefined' ? 40 : modelOptions.topK, // 1-40, default: 40
-      stop: modelOptions.stop, // no stop method for now
-    };
+    this.modelOptions = Object.assign(
+      {
+        model: anthropicSettings.model.default,
+      },
+      this.modelOptions,
+      this.options.modelOptions,
+    );
+
+    const modelMatch = matchModelName(this.modelOptions.model, EModelEndpoint.anthropic);
+    this.isClaude3 = modelMatch.startsWith('claude-3');
+    this.isLegacyOutput = !modelMatch.startsWith('claude-3-5-sonnet');
+    this.supportsCacheControl =
+      this.options.promptCache && this.checkPromptCacheSupport(modelMatch);
+
+    if (
+      this.isLegacyOutput &&
+      this.modelOptions.maxOutputTokens &&
+      this.modelOptions.maxOutputTokens > legacy.maxOutputTokens.default
+    ) {
+      this.modelOptions.maxOutputTokens = legacy.maxOutputTokens.default;
+    }
 
-    this.isClaude3 = this.modelOptions.model.includes('claude-3');
     this.useMessages = this.isClaude3 || !!this.options.attachments;
 
     this.defaultVisionModel = this.options.visionModel ?? 'claude-3-sonnet-20240229';
@@ -118,25 +154,92 @@ class AnthropicClient extends BaseClient {
 
   /**
    * Get the initialized Anthropic client.
+   * @param {Partial<Anthropic.ClientOptions>} requestOptions - The options for the client.
    * @returns {Anthropic} The Anthropic client instance.
    */
-  getClient() {
-    /** @type {Anthropic.default.RequestOptions} */
+  getClient(requestOptions) {
+    /** @type {Anthropic.ClientOptions} */
     const options = {
+      fetch: this.fetch,
       apiKey: this.apiKey,
     };
 
+    if (this.options.proxy) {
+      options.httpAgent = new HttpsProxyAgent(this.options.proxy);
+    }
+
     if (this.options.reverseProxyUrl) {
       options.baseURL = this.options.reverseProxyUrl;
     }
 
+    if (
+      this.supportsCacheControl &&
+      requestOptions?.model &&
+      requestOptions.model.includes('claude-3-5-sonnet')
+    ) {
+      options.defaultHeaders = {
+        'anthropic-beta': 'max-tokens-3-5-sonnet-2024-07-15,prompt-caching-2024-07-31',
+      };
+    } else if (this.supportsCacheControl) {
+      options.defaultHeaders = {
+        'anthropic-beta': 'prompt-caching-2024-07-31',
+      };
+    }
+
     return new Anthropic(options);
   }
 
-  getTokenCountForResponse(response) {
+  /**
+   * Get stream usage as returned by this client's API response.
+   * @returns {AnthropicStreamUsage} The stream usage object.
+   */
+  getStreamUsage() {
+    const inputUsage = this.message_start?.message?.usage ?? {};
+    const outputUsage = this.message_delta?.usage ?? {};
+    return Object.assign({}, inputUsage, outputUsage);
+  }
+
+  /**
+   * Calculates the correct token count for the current message based on the token count map and API usage.
+   * Edge case: If the calculation results in a negative value, it returns the original estimate.
+   * If revisiting a conversation with a chat history entirely composed of token estimates,
+   * the cumulative token count going forward should become more accurate as the conversation progresses.
+   * @param {Object} params - The parameters for the calculation.
+   * @param {Record<string, number>} params.tokenCountMap - A map of message IDs to their token counts.
+   * @param {string} params.currentMessageId - The ID of the current message to calculate.
+   * @param {AnthropicStreamUsage} params.usage - The usage object returned by the API.
+   * @returns {number} The correct token count for the current message.
+   */
+  calculateCurrentTokenCount({ tokenCountMap, currentMessageId, usage }) {
+    const originalEstimate = tokenCountMap[currentMessageId] || 0;
+
+    if (!usage || typeof usage.input_tokens !== 'number') {
+      return originalEstimate;
+    }
+
+    tokenCountMap[currentMessageId] = 0;
+    const totalTokensFromMap = Object.values(tokenCountMap).reduce((sum, count) => {
+      const numCount = Number(count);
+      return sum + (isNaN(numCount) ? 0 : numCount);
+    }, 0);
+    const totalInputTokens =
+      (usage.input_tokens ?? 0) +
+      (usage.cache_creation_input_tokens ?? 0) +
+      (usage.cache_read_input_tokens ?? 0);
+
+    const currentMessageTokens = totalInputTokens - totalTokensFromMap;
+    return currentMessageTokens > 0 ? currentMessageTokens : originalEstimate;
+  }
+
+  /**
+   * Get Token Count for LibreChat Message
+   * @param {TMessage} responseMessage
+   * @returns {number}
+   */
+  getTokenCountForResponse(responseMessage) {
     return this.getTokenCountForMessage({
       role: 'assistant',
-      content: response.text,
+      content: responseMessage.text,
     });
   }
 
@@ -189,7 +292,38 @@ class AnthropicClient extends BaseClient {
     return files;
   }
 
-  async recordTokenUsage({ promptTokens, completionTokens, model, context = 'message' }) {
+  /**
+   * @param {object} params
+   * @param {number} params.promptTokens
+   * @param {number} params.completionTokens
+   * @param {AnthropicStreamUsage} [params.usage]
+   * @param {string} [params.model]
+   * @param {string} [params.context='message']
+   * @returns {Promise<void>}
+   */
+  async recordTokenUsage({ promptTokens, completionTokens, usage, model, context = 'message' }) {
+    if (usage != null && usage?.input_tokens != null) {
+      const input = usage.input_tokens ?? 0;
+      const write = usage.cache_creation_input_tokens ?? 0;
+      const read = usage.cache_read_input_tokens ?? 0;
+
+      await spendStructuredTokens(
+        {
+          context,
+          user: this.user,
+          conversationId: this.conversationId,
+          model: model ?? this.modelOptions.model,
+          endpointTokenConfig: this.options.endpointTokenConfig,
+        },
+        {
+          promptTokens: { input, write, read },
+          completionTokens,
+        },
+      );
+
+      return;
+    }
+
     await spendTokens(
       {
         context,
@@ -537,6 +671,18 @@ class AnthropicClient extends BaseClient {
       : await client.completions.create(options);
   }
 
+  /**
+   * @param {string} modelName
+   * @returns {boolean}
+   */
+  checkPromptCacheSupport(modelName) {
+    const modelMatch = matchModelName(modelName, EModelEndpoint.anthropic);
+    if (modelMatch === 'claude-3-5-sonnet' || modelMatch === 'claude-3-haiku') {
+      return true;
+    }
+    return false;
+  }
+
   async sendCompletion(payload, { onProgress, abortController }) {
     if (!abortController) {
       abortController = new AbortController();
@@ -550,8 +696,6 @@ class AnthropicClient extends BaseClient {
     }
 
     logger.debug('modelOptions', { modelOptions });
-
-    const client = this.getClient();
     const metadata = {
       user_id: this.user,
     };
@@ -579,16 +723,28 @@ class AnthropicClient extends BaseClient {
 
     if (this.useMessages) {
       requestOptions.messages = payload;
-      requestOptions.max_tokens = maxOutputTokens || 1500;
+      requestOptions.max_tokens = maxOutputTokens || legacy.maxOutputTokens.default;
     } else {
       requestOptions.prompt = payload;
       requestOptions.max_tokens_to_sample = maxOutputTokens || 1500;
     }
 
-    if (this.systemMessage) {
+    if (this.systemMessage && this.supportsCacheControl === true) {
+      requestOptions.system = [
+        {
+          type: 'text',
+          text: this.systemMessage,
+          cache_control: { type: 'ephemeral' },
+        },
+      ];
+    } else if (this.systemMessage) {
       requestOptions.system = this.systemMessage;
     }
 
+    if (this.supportsCacheControl === true && this.useMessages) {
+      requestOptions.messages = addCacheControl(requestOptions.messages);
+    }
+
     logger.debug('[AnthropicClient]', { ...requestOptions });
 
     const handleChunk = (currentChunk) => {
@@ -599,12 +755,14 @@ class AnthropicClient extends BaseClient {
     };
 
     const maxRetries = 3;
+    const streamRate = this.options.streamRate ?? Constants.DEFAULT_STREAM_RATE;
     async function processResponse() {
       let attempts = 0;
 
       while (attempts < maxRetries) {
         let response;
         try {
+          const client = this.getClient(requestOptions);
           response = await this.createResponse(client, requestOptions);
 
           signal.addEventListener('abort', () => {
@@ -616,11 +774,18 @@ class AnthropicClient extends BaseClient {
 
           for await (const completion of response) {
             // Handle each completion as before
+            const type = completion?.type ?? '';
+            if (tokenEventTypes.has(type)) {
+              logger.debug(`[AnthropicClient] ${type}`, completion);
+              this[type] = completion;
+            }
             if (completion?.delta?.text) {
               handleChunk(completion.delta.text);
             } else if (completion.completion) {
               handleChunk(completion.completion);
             }
+
+            await sleep(streamRate);
           }
 
           // Successful processing, exit loop
@@ -657,6 +822,7 @@ class AnthropicClient extends BaseClient {
       maxContextTokens: this.options.maxContextTokens,
       promptPrefix: this.options.promptPrefix,
       modelLabel: this.options.modelLabel,
+      promptCache: this.options.promptCache,
       resendFiles: this.options.resendFiles,
       iconURL: this.options.iconURL,
       greeting: this.options.greeting,
@@ -702,6 +868,8 @@ class AnthropicClient extends BaseClient {
    */
   async titleConvo({ text, responseText = '' }) {
     let title = 'New Chat';
+    this.message_delta = undefined;
+    this.message_start = undefined;
     const convo = `<initial_message>
   ${truncateText(text)}
   </initial_message>
@@ -731,7 +899,11 @@ class AnthropicClient extends BaseClient {
       };
 
       try {
-        const response = await this.createResponse(this.getClient(), requestOptions, true);
+        const response = await this.createResponse(
+          this.getClient(requestOptions),
+          requestOptions,
+          true,
+        );
         let promptTokens = response?.usage?.input_tokens;
         let completionTokens = response?.usage?.output_tokens;
         if (!promptTokens) {

api/app/clients/BaseClient.js

@@ -1,9 +1,11 @@
 const crypto = require('crypto');
-const { supportsBalanceCheck, Constants } = require('librechat-data-provider');
-const { getConvo, getMessages, saveMessage, updateMessage, saveConvo } = require('~/models');
+const fetch = require('node-fetch');
+const { supportsBalanceCheck, Constants, CacheKeys, Time } = require('librechat-data-provider');
+const { getMessages, saveMessage, updateMessage, saveConvo } = require('~/models');
 const { addSpaceIfNeeded, isEnabled } = require('~/server/utils');
 const checkBalance = require('~/models/checkBalance');
 const { getFiles } = require('~/models/File');
+const { getLogStores } = require('~/cache');
 const TextStream = require('./TextStream');
 const { logger } = require('~/config');
 
@@ -17,6 +19,15 @@ class BaseClient {
       month: 'long',
       day: 'numeric',
     });
+    this.fetch = this.fetch.bind(this);
+    /** @type {boolean} */
+    this.skipSaveConvo = false;
+    /** @type {boolean} */
+    this.skipSaveUserMessage = false;
+    /** @type {ClientDatabaseSavePromise} */
+    this.userMessagePromise;
+    /** @type {ClientDatabaseSavePromise} */
+    this.responsePromise;
   }
 
   setOptions() {
@@ -43,10 +54,22 @@ class BaseClient {
     throw new Error('Subclasses attempted to call summarizeMessages without implementing it');
   }
 
-  async getTokenCountForResponse(response) {
-    logger.debug('`[BaseClient] recordTokenUsage` not implemented.', response);
+  /**
+   * Abstract method to get the token count for a message. Subclasses must implement this method.
+   * @param {TMessage} responseMessage
+   * @returns {number}
+   */
+  getTokenCountForResponse(responseMessage) {
+    logger.debug('`[BaseClient] recordTokenUsage` not implemented.', responseMessage);
   }
 
+  /**
+   * Abstract method to record token usage. Subclasses must implement this method.
+   * If a correction to the token usage is needed, the method should return an object with the corrected token counts.
+   * @param {number} promptTokens
+   * @param {number} completionTokens
+   * @returns {Promise<void>}
+   */
   async recordTokenUsage({ promptTokens, completionTokens }) {
     logger.debug('`[BaseClient] recordTokenUsage` not implemented.', {
       promptTokens,
@@ -54,6 +77,25 @@ class BaseClient {
     });
   }
 
+  /**
+   * Makes an HTTP request and logs the process.
+   *
+   * @param {RequestInfo} url - The URL to make the request to. Can be a string or a Request object.
+   * @param {RequestInit} [init] - Optional init options for the request.
+   * @returns {Promise<Response>} - A promise that resolves to the response of the fetch request.
+   */
+  async fetch(_url, init) {
+    let url = _url;
+    if (this.options.directEndpoint) {
+      url = this.options.reverseProxyUrl;
+    }
+    logger.debug(`Making request to ${url}`);
+    if (typeof Bun !== 'undefined') {
+      return await fetch(url, init);
+    }
+    return await fetch(url, init);
+  }
+
   getBuildMessagesOptions() {
     throw new Error('Subclasses must implement getBuildMessagesOptions');
   }
@@ -63,19 +105,45 @@ class BaseClient {
     await stream.processTextStream(onProgress);
   }
 
+  /**
+   * @returns {[string|undefined, string|undefined]}
+   */
+  processOverideIds() {
+    /** @type {Record<string, string | undefined>} */
+    let { overrideConvoId, overrideUserMessageId } = this.options?.req?.body ?? {};
+    if (overrideConvoId) {
+      const [conversationId, index] = overrideConvoId.split(Constants.COMMON_DIVIDER);
+      overrideConvoId = conversationId;
+      if (index !== '0') {
+        this.skipSaveConvo = true;
+      }
+    }
+    if (overrideUserMessageId) {
+      const [userMessageId, index] = overrideUserMessageId.split(Constants.COMMON_DIVIDER);
+      overrideUserMessageId = userMessageId;
+      if (index !== '0') {
+        this.skipSaveUserMessage = true;
+      }
+    }
+
+    return [overrideConvoId, overrideUserMessageId];
+  }
+
   async setMessageOptions(opts = {}) {
     if (opts && opts.replaceOptions) {
       this.setOptions(opts);
     }
 
+    const [overrideConvoId, overrideUserMessageId] = this.processOverideIds();
     const { isEdited, isContinued } = opts;
     const user = opts.user ?? null;
     this.user = user;
     const saveOptions = this.getSaveOptions();
     this.abortController = opts.abortController ?? new AbortController();
-    const conversationId = opts.conversationId ?? crypto.randomUUID();
+    const conversationId = overrideConvoId ?? opts.conversationId ?? crypto.randomUUID();
     const parentMessageId = opts.parentMessageId ?? Constants.NO_PARENT;
-    const userMessageId = opts.overrideParentMessageId ?? crypto.randomUUID();
+    const userMessageId =
+      overrideUserMessageId ?? opts.overrideParentMessageId ?? crypto.randomUUID();
     let responseMessageId = opts.responseMessageId ?? crypto.randomUUID();
     let head = isEdited ? responseMessageId : parentMessageId;
     this.currentMessages = (await this.loadHistory(conversationId, head)) ?? [];
@@ -139,7 +207,7 @@ class BaseClient {
     }
 
     if (typeof opts?.onStart === 'function') {
-      opts.onStart(userMessage);
+      opts.onStart(userMessage, responseMessageId);
     }
 
     return {
@@ -373,6 +441,14 @@ class BaseClient {
     const { user, head, isEdited, conversationId, responseMessageId, saveOptions, userMessage } =
       await this.handleStartMethods(message, opts);
 
+    if (opts.progressCallback) {
+      opts.onProgress = opts.progressCallback.call(null, {
+        ...(opts.progressOptions ?? {}),
+        parentMessageId: userMessage.messageId,
+        messageId: responseMessageId,
+      });
+    }
+
     const { generation = '' } = opts;
 
     // It's not necessary to push to currentMessages
@@ -421,8 +497,13 @@ class BaseClient {
       this.handleTokenCountMap(tokenCountMap);
     }
 
-    if (!isEdited) {
-      await this.saveMessageToDatabase(userMessage, saveOptions, user);
+    if (!isEdited && !this.skipSaveUserMessage) {
+      this.userMessagePromise = this.saveMessageToDatabase(userMessage, saveOptions, user);
+      if (typeof opts?.getReqData === 'function') {
+        opts.getReqData({
+          userMessagePromise: this.userMessagePromise,
+        });
+      }
     }
 
     if (
@@ -467,17 +548,103 @@ class BaseClient {
       this.getTokenCountForResponse &&
       this.getTokenCount
     ) {
-      responseMessage.tokenCount = this.getTokenCountForResponse(responseMessage);
-      const completionTokens = this.getTokenCount(completion);
-      await this.recordTokenUsage({ promptTokens, completionTokens });
+      let completionTokens;
+
+      /**
+       * Metadata about input/output costs for the current message. The client
+       * should provide a function to get the current stream usage metadata; if not,
+       * use the legacy token estimations.
+       * @type {StreamUsage | null} */
+      const usage = this.getStreamUsage != null ? this.getStreamUsage() : null;
+
+      if (usage != null && Number(usage.output_tokens) > 0) {
+        responseMessage.tokenCount = usage.output_tokens;
+        completionTokens = responseMessage.tokenCount;
+        await this.updateUserMessageTokenCount({ usage, tokenCountMap, userMessage, opts });
+      } else {
+        responseMessage.tokenCount = this.getTokenCountForResponse(responseMessage);
+        completionTokens = this.getTokenCount(completion);
+      }
+
+      await this.recordTokenUsage({ promptTokens, completionTokens, usage });
     }
-    await this.saveMessageToDatabase(responseMessage, saveOptions, user);
+
+    if (this.userMessagePromise) {
+      await this.userMessagePromise;
+    }
+
+    this.responsePromise = this.saveMessageToDatabase(responseMessage, saveOptions, user);
+    const messageCache = getLogStores(CacheKeys.MESSAGES);
+    messageCache.set(
+      responseMessageId,
+      {
+        text: responseMessage.text,
+        complete: true,
+      },
+      Time.FIVE_MINUTES,
+    );
     delete responseMessage.tokenCount;
     return responseMessage;
   }
 
-  async getConversation(conversationId, user = null) {
-    return await getConvo(user, conversationId);
+  /**
+   * Stream usage should only be used for user message token count re-calculation if:
+   * - The stream usage is available, with input tokens greater than 0,
+   * - the client provides a function to calculate the current token count,
+   * - files are being resent with every message (default behavior; or if `false`, with no attachments),
+   * - the `promptPrefix` (custom instructions) is not set.
+   *
+   * In these cases, the legacy token estimations would be more accurate.
+   *
+   * TODO: included system messages in the `orderedMessages` accounting, potentially as a
+   * separate message in the UI. ChatGPT does this through "hidden" system messages.
+   * @param {object} params
+   * @param {StreamUsage} params.usage
+   * @param {Record<string, number>} params.tokenCountMap
+   * @param {TMessage} params.userMessage
+   * @param {object} params.opts
+   */
+  async updateUserMessageTokenCount({ usage, tokenCountMap, userMessage, opts }) {
+    /** @type {boolean} */
+    const shouldUpdateCount =
+      this.calculateCurrentTokenCount != null &&
+      Number(usage.input_tokens) > 0 &&
+      (this.options.resendFiles ||
+        (!this.options.resendFiles && !this.options.attachments?.length)) &&
+      !this.options.promptPrefix;
+
+    if (!shouldUpdateCount) {
+      return;
+    }
+
+    const userMessageTokenCount = this.calculateCurrentTokenCount({
+      currentMessageId: userMessage.messageId,
+      tokenCountMap,
+      usage,
+    });
+
+    if (userMessageTokenCount === userMessage.tokenCount) {
+      return;
+    }
+
+    userMessage.tokenCount = userMessageTokenCount;
+    /*
+      Note: `AskController` saves the user message, so we update the count of its `userMessage` reference
+    */
+    if (typeof opts?.getReqData === 'function') {
+      opts.getReqData({
+        userMessage,
+      });
+    }
+    /*
+      Note: we update the user message to be sure it gets the calculated token count;
+      though `AskController` saves the user message, EditController does not
+    */
+    await this.userMessagePromise;
+    await this.updateMessageInDatabase({
+      messageId: userMessage.messageId,
+      tokenCount: userMessageTokenCount,
+    });
   }
 
   async loadHistory(conversationId, parentMessageId = null) {
@@ -534,22 +701,45 @@ class BaseClient {
    * @param {string | null} user
    */
   async saveMessageToDatabase(message, endpointOptions, user = null) {
-    await saveMessage({
-      ...message,
-      endpoint: this.options.endpoint,
-      unfinished: false,
-      user,
-    });
-    await saveConvo(user, {
-      conversationId: message.conversationId,
-      endpoint: this.options.endpoint,
-      endpointType: this.options.endpointType,
-      ...endpointOptions,
-    });
+    if (this.user && user !== this.user) {
+      throw new Error('User mismatch.');
+    }
+
+    const savedMessage = await saveMessage(
+      this.options.req,
+      {
+        ...message,
+        endpoint: this.options.endpoint,
+        unfinished: false,
+        user,
+      },
+      { context: 'api/app/clients/BaseClient.js - saveMessageToDatabase #saveMessage' },
+    );
+
+    if (this.skipSaveConvo) {
+      return { message: savedMessage };
+    }
+
+    const conversation = await saveConvo(
+      this.options.req,
+      {
+        conversationId: message.conversationId,
+        endpoint: this.options.endpoint,
+        endpointType: this.options.endpointType,
+        ...endpointOptions,
+      },
+      { context: 'api/app/clients/BaseClient.js - saveMessageToDatabase #saveConvo' },
+    );
+
+    return { message: savedMessage, conversation };
   }
 
+  /**
+   * Update a message in the database.
+   * @param {Partial<TMessage>} message
+   */
   async updateMessageInDatabase(message) {
-    await updateMessage(message);
+    await updateMessage(this.options.req, message);
   }
 
   /**

api/app/clients/ChatGPTClient.js

@@ -438,9 +438,17 @@ class ChatGPTClient extends BaseClient {
 
       if (message.eventType === 'text-generation' && message.text) {
         onTokenProgress(message.text);
-      } else if (message.eventType === 'stream-end' && message.response) {
+        reply += message.text;
+      }
+      /*
+      Cohere API Chinese Unicode character replacement hotfix.
+      Should be un-commented when the following issue is resolved:
+      https://github.com/cohere-ai/cohere-typescript/issues/151
+
+      else if (message.eventType === 'stream-end' && message.response) {
         reply = message.response.text;
       }
+      */
     }
 
     return reply;

api/app/clients/GoogleClient.js

@@ -13,13 +13,20 @@ const {
   endpointSettings,
   EModelEndpoint,
   VisionModes,
+  Constants,
   AuthKeys,
 } = require('librechat-data-provider');
 const { encodeAndFormat } = require('~/server/services/Files/images');
-const { formatMessage, createContextHandlers } = require('./prompts');
 const { getModelMaxTokens } = require('~/utils');
-const BaseClient = require('./BaseClient');
+const { sleep } = require('~/server/utils');
 const { logger } = require('~/config');
+const {
+  formatMessage,
+  createContextHandlers,
+  titleInstruction,
+  truncateText,
+} = require('./prompts');
+const BaseClient = require('./BaseClient');
 
 const loc = 'us-central1';
 const publisher = 'google';
@@ -113,19 +120,7 @@ class GoogleClient extends BaseClient {
       .filter((ex) => ex)
       .filter((obj) => obj.input.content !== '' && obj.output.content !== '');
 
-    const modelOptions = this.options.modelOptions || {};
-    this.modelOptions = {
-      ...modelOptions,
-      // set some good defaults (check for undefined in some cases because they may be 0)
-      model: modelOptions.model || settings.model.default,
-      temperature:
-        typeof modelOptions.temperature === 'undefined'
-          ? settings.temperature.default
-          : modelOptions.temperature,
-      topP: typeof modelOptions.topP === 'undefined' ? settings.topP.default : modelOptions.topP,
-      topK: typeof modelOptions.topK === 'undefined' ? settings.topK.default : modelOptions.topK,
-      // stop: modelOptions.stop // no stop method for now
-    };
+    this.modelOptions = this.options.modelOptions || {};
 
     this.options.attachments?.then((attachments) => this.checkVisionRequest(attachments));
 
@@ -591,12 +586,16 @@ class GoogleClient extends BaseClient {
   createLLM(clientOptions) {
     const model = clientOptions.modelName ?? clientOptions.model;
     if (this.project_id && this.isTextModel) {
+      logger.debug('Creating Google VertexAI client');
       return new GoogleVertexAI(clientOptions);
     } else if (this.project_id && this.isChatModel) {
+      logger.debug('Creating Chat Google VertexAI client');
       return new ChatGoogleVertexAI(clientOptions);
     } else if (this.project_id) {
+      logger.debug('Creating VertexAI client');
       return new ChatVertexAI(clientOptions);
     } else if (model.includes('1.5')) {
+      logger.debug('Creating GenAI client');
       return new GenAI(this.apiKey).getGenerativeModel(
         {
           ...clientOptions,
@@ -606,12 +605,14 @@ class GoogleClient extends BaseClient {
       );
     }
 
+    logger.debug('Creating Chat Google Generative AI client');
     return new ChatGoogleGenerativeAI({ ...clientOptions, apiKey: this.apiKey });
   }
 
   async getCompletion(_payload, options = {}) {
-    const { onProgress, abortController } = options;
     const { parameters, instances } = _payload;
+    const { onProgress, abortController } = options;
+    const streamRate = this.options.streamRate ?? Constants.DEFAULT_STREAM_RATE;
     const { messages: _messages, context, examples: _examples } = instances?.[0] ?? {};
 
     let examples;
@@ -664,6 +665,102 @@ class GoogleClient extends BaseClient {
 
     const modelName = clientOptions.modelName ?? clientOptions.model ?? '';
     if (modelName?.includes('1.5') && !this.project_id) {
+      const client = model;
+      const requestOptions = {
+        contents: _payload,
+      };
+
+      if (this.options?.promptPrefix?.length) {
+        requestOptions.systemInstruction = {
+          parts: [
+            {
+              text: this.options.promptPrefix,
+            },
+          ],
+        };
+      }
+
+      requestOptions.safetySettings = _payload.safetySettings;
+
+      const delay = modelName.includes('flash') ? 8 : 14;
+      const result = await client.generateContentStream(requestOptions);
+      for await (const chunk of result.stream) {
+        const chunkText = chunk.text();
+        await this.generateTextStream(chunkText, onProgress, {
+          delay,
+        });
+        reply += chunkText;
+        await sleep(streamRate);
+      }
+      return reply;
+    }
+
+    const stream = await model.stream(messages, {
+      signal: abortController.signal,
+      timeout: 7000,
+      safetySettings: _payload.safetySettings,
+    });
+
+    let delay = this.options.streamRate || 8;
+
+    if (!this.options.streamRate) {
+      if (this.isGenerativeModel) {
+        delay = 12;
+      }
+      if (modelName.includes('flash')) {
+        delay = 5;
+      }
+    }
+
+    for await (const chunk of stream) {
+      const chunkText = chunk?.content ?? chunk;
+      await this.generateTextStream(chunkText, onProgress, {
+        delay,
+      });
+      reply += chunkText;
+    }
+
+    return reply;
+  }
+
+  /**
+   * Stripped-down logic for generating a title. This uses the non-streaming APIs, since the user does not see titles streaming
+   */
+  async titleChatCompletion(_payload, options = {}) {
+    const { abortController } = options;
+    const { parameters, instances } = _payload;
+    const { messages: _messages, examples: _examples } = instances?.[0] ?? {};
+
+    let clientOptions = { ...parameters, maxRetries: 2 };
+
+    logger.debug('Initialized title client options');
+
+    if (this.project_id) {
+      clientOptions['authOptions'] = {
+        credentials: {
+          ...this.serviceKey,
+        },
+        projectId: this.project_id,
+      };
+    }
+
+    if (!parameters) {
+      clientOptions = { ...clientOptions, ...this.modelOptions };
+    }
+
+    if (this.isGenerativeModel && !this.project_id) {
+      clientOptions.modelName = clientOptions.model;
+      delete clientOptions.model;
+    }
+
+    const model = this.createLLM(clientOptions);
+
+    let reply = '';
+    const messages = this.isTextModel ? _payload.trim() : _messages;
+
+    const modelName = clientOptions.modelName ?? clientOptions.model ?? '';
+    if (modelName?.includes('1.5') && !this.project_id) {
+      logger.debug('Identified titling model as 1.5 version');
       /** @type {GenerativeModel} */
       const client = model;
       const requestOptions = {
@@ -683,38 +780,64 @@ class GoogleClient extends BaseClient {
       const safetySettings = _payload.safetySettings;
       requestOptions.safetySettings = safetySettings;
 
-      const delay = modelName.includes('flash') ? 8 : 14;
-      const result = await client.generateContentStream(requestOptions);
-      for await (const chunk of result.stream) {
-        const chunkText = chunk.text();
-        await this.generateTextStream(chunkText, onProgress, {
-          delay,
-        });
-        reply += chunkText;
-      }
+      const result = await client.generateContent(requestOptions);
+
+      reply = result.response?.text();
+
+      return reply;
+    } else {
+      logger.debug('Beginning titling');
+      const safetySettings = _payload.safetySettings;
+
+      const titleResponse = await model.invoke(messages, {
+        signal: abortController.signal,
+        timeout: 7000,
+        safetySettings: safetySettings,
+      });
+
+      reply = titleResponse.content;
+      // TODO: RECORD TOKEN USAGE
       return reply;
     }
+  }
 
-    const safetySettings = _payload.safetySettings;
-    const stream = await model.stream(messages, {
-      signal: abortController.signal,
-      timeout: 7000,
-      safetySettings: safetySettings,
-    });
+  async titleConvo({ text, responseText = '' }) {
+    let title = 'New Chat';
+    const convo = `||>User:
+"${truncateText(text)}"
+||>Response:
+"${JSON.stringify(truncateText(responseText))}"`;
 
-    let delay = this.isGenerativeModel ? 12 : 8;
-    if (modelName.includes('flash')) {
-      delay = 5;
+    let { prompt: payload } = await this.buildMessages([
+      {
+        text: `Please generate ${titleInstruction}
+
+    ${convo}
+    
+    ||>Title:`,
+        isCreatedByUser: true,
+        author: this.userLabel,
+      },
+    ]);
+
+    if (this.isVisionModel) {
+      logger.warn(
+        `Current vision model does not support titling without an attachment; falling back to default model ${settings.model.default}`,
+      );
+
+      payload.parameters = { ...payload.parameters, model: settings.model.default };
     }
-    for await (const chunk of stream) {
-      const chunkText = chunk?.content ?? chunk;
-      await this.generateTextStream(chunkText, onProgress, {
-        delay,
+
+    try {
+      title = await this.titleChatCompletion(payload, {
+        abortController: new AbortController(),
+        onProgress: () => {},
       });
-      reply += chunkText;
+    } catch (e) {
+      logger.error('[GoogleClient] There was an issue generating the title', e);
     }
-
-    return reply;
+    logger.debug(`Title response: ${title}`);
+    return title;
   }
 
   getSaveOptions() {
@@ -733,38 +856,36 @@ class GoogleClient extends BaseClient {
   }
 
   async sendCompletion(payload, opts = {}) {
-    const modelName = payload.parameters?.model;
-
-    if (modelName && modelName.toLowerCase().includes('gemini')) {
-      const safetySettings = [
-        {
-          category: 'HARM_CATEGORY_SEXUALLY_EXPLICIT',
-          threshold:
-            process.env.GOOGLE_SAFETY_SEXUALLY_EXPLICIT || 'HARM_BLOCK_THRESHOLD_UNSPECIFIED',
-        },
-        {
-          category: 'HARM_CATEGORY_HATE_SPEECH',
-          threshold: process.env.GOOGLE_SAFETY_HATE_SPEECH || 'HARM_BLOCK_THRESHOLD_UNSPECIFIED',
-        },
-        {
-          category: 'HARM_CATEGORY_HARASSMENT',
-          threshold: process.env.GOOGLE_SAFETY_HARASSMENT || 'HARM_BLOCK_THRESHOLD_UNSPECIFIED',
-        },
-        {
-          category: 'HARM_CATEGORY_DANGEROUS_CONTENT',
-          threshold:
-            process.env.GOOGLE_SAFETY_DANGEROUS_CONTENT || 'HARM_BLOCK_THRESHOLD_UNSPECIFIED',
-        },
-      ];
-
-      payload.safetySettings = safetySettings;
-    }
+    payload.safetySettings = this.getSafetySettings();
 
     let reply = '';
     reply = await this.getCompletion(payload, opts);
     return reply.trim();
   }
 
+  getSafetySettings() {
+    return [
+      {
+        category: 'HARM_CATEGORY_SEXUALLY_EXPLICIT',
+        threshold:
+          process.env.GOOGLE_SAFETY_SEXUALLY_EXPLICIT || 'HARM_BLOCK_THRESHOLD_UNSPECIFIED',
+      },
+      {
+        category: 'HARM_CATEGORY_HATE_SPEECH',
+        threshold: process.env.GOOGLE_SAFETY_HATE_SPEECH || 'HARM_BLOCK_THRESHOLD_UNSPECIFIED',
+      },
+      {
+        category: 'HARM_CATEGORY_HARASSMENT',
+        threshold: process.env.GOOGLE_SAFETY_HARASSMENT || 'HARM_BLOCK_THRESHOLD_UNSPECIFIED',
+      },
+      {
+        category: 'HARM_CATEGORY_DANGEROUS_CONTENT',
+        threshold:
+          process.env.GOOGLE_SAFETY_DANGEROUS_CONTENT || 'HARM_BLOCK_THRESHOLD_UNSPECIFIED',
+      },
+    ];
+  }
+
   /* TO-DO: Handle tokens with Google tokenization NOTE: these are required */
   static getTokenizer(encoding, isModelName = false, extendSpecialTokens = {}) {
     if (tokenizersCache[encoding]) {

api/app/clients/OllamaClient.js

@@ -1,7 +1,9 @@
 const { z } = require('zod');
 const axios = require('axios');
 const { Ollama } = require('ollama');
+const { Constants } = require('librechat-data-provider');
 const { deriveBaseURL } = require('~/utils');
+const { sleep } = require('~/server/utils');
 const { logger } = require('~/config');
 
 const ollamaPayloadSchema = z.object({
@@ -40,6 +42,7 @@ const getValidBase64 = (imageUrl) => {
 class OllamaClient {
   constructor(options = {}) {
     const host = deriveBaseURL(options.baseURL ?? 'http://localhost:11434');
+    this.streamRate = options.streamRate ?? Constants.DEFAULT_STREAM_RATE;
     /** @type {Ollama} */
     this.client = new Ollama({ host });
   }
@@ -136,6 +139,8 @@ class OllamaClient {
           stream.controller.abort();
           break;
         }
+
+        await sleep(this.streamRate);
       }
     }
     // TODO: regular completion

api/app/clients/OpenAIClient.js

@@ -6,6 +6,7 @@ const {
   ImageDetail,
   EModelEndpoint,
   resolveHeaders,
+  openAISettings,
   ImageDetailCost,
   CohereConstants,
   getResponseSender,
@@ -27,9 +28,9 @@ const {
   createContextHandlers,
 } = require('./prompts');
 const { encodeAndFormat } = require('~/server/services/Files/images/encode');
+const { spendTokens } = require('~/models/spendTokens');
 const { isEnabled, sleep } = require('~/server/utils');
 const { handleOpenAIErrors } = require('./tools/util');
-const spendTokens = require('~/models/spendTokens');
 const { createLLM, RunManager } = require('./llm');
 const ChatGPTClient = require('./ChatGPTClient');
 const { summaryBuffer } = require('./memory');
@@ -85,26 +86,13 @@ class OpenAIClient extends BaseClient {
       this.apiKey = this.options.openaiApiKey;
     }
 
-    const modelOptions = this.options.modelOptions || {};
-
-    if (!this.modelOptions) {
-      this.modelOptions = {
-        ...modelOptions,
-        model: modelOptions.model || 'gpt-3.5-turbo',
-        temperature:
-          typeof modelOptions.temperature === 'undefined' ? 0.8 : modelOptions.temperature,
-        top_p: typeof modelOptions.top_p === 'undefined' ? 1 : modelOptions.top_p,
-        presence_penalty:
-          typeof modelOptions.presence_penalty === 'undefined' ? 1 : modelOptions.presence_penalty,
-        stop: modelOptions.stop,
-      };
-    } else {
-      // Update the modelOptions if it already exists
-      this.modelOptions = {
-        ...this.modelOptions,
-        ...modelOptions,
-      };
-    }
+    this.modelOptions = Object.assign(
+      {
+        model: openAISettings.model.default,
+      },
+      this.modelOptions,
+      this.options.modelOptions,
+    );
 
     this.defaultVisionModel = this.options.visionModel ?? 'gpt-4-vision-preview';
     if (typeof this.options.attachments?.then === 'function') {
@@ -588,7 +576,7 @@ class OpenAIClient extends BaseClient {
     let streamResult = null;
     this.modelOptions.user = this.user;
     const invalidBaseUrl = this.completionsUrl && extractBaseURL(this.completionsUrl) === null;
-    const useOldMethod = !!(invalidBaseUrl || !this.isChatCompletion || typeof Bun !== 'undefined');
+    const useOldMethod = !!(invalidBaseUrl || !this.isChatCompletion);
     if (typeof opts.onProgress === 'function' && useOldMethod) {
       const completionResult = await this.getCompletion(
         payload,
@@ -756,6 +744,8 @@ class OpenAIClient extends BaseClient {
    *                            In case of failure, it will return the default title, "New Chat".
    */
   async titleConvo({ text, conversationId, responseText = '' }) {
+    this.conversationId = conversationId;
+
     if (this.options.attachments) {
       delete this.options.attachments;
     }
@@ -825,7 +815,7 @@ class OpenAIClient extends BaseClient {
 
       const instructionsPayload = [
         {
-          role: 'system',
+          role: this.options.titleMessageRole ?? (this.isOllama ? 'user' : 'system'),
           content: `Please generate ${titleInstruction}
 
 ${convo}
@@ -838,13 +828,17 @@ ${convo}
 
       try {
         let useChatCompletion = true;
+
         if (this.options.reverseProxyUrl === CohereConstants.API_URL) {
           useChatCompletion = false;
         }
+
         title = (
           await this.sendPayload(instructionsPayload, { modelOptions, useChatCompletion })
         ).replaceAll('"', '');
+
         const completionTokens = this.getTokenCount(title);
+
         this.recordTokenUsage({ promptTokens, completionTokens, context: 'title' });
       } catch (e) {
         logger.error(
@@ -868,6 +862,7 @@ ${convo}
         context: 'title',
         tokenBuffer: 150,
       });
+
       title = await runTitleChain({ llm, text, convo, signal: this.abortController.signal });
     } catch (e) {
       if (e?.message?.toLowerCase()?.includes('abort')) {
@@ -1005,9 +1000,9 @@ ${convo}
     await spendTokens(
       {
         context,
-        user: this.user,
         model: this.modelOptions.model,
         conversationId: this.conversationId,
+        user: this.user ?? this.options.req.user?.id,
         endpointTokenConfig: this.options.endpointTokenConfig,
       },
       { promptTokens, completionTokens },
@@ -1099,7 +1094,12 @@ ${convo}
       }
 
       if (this.azure || this.options.azure) {
-        // Azure does not accept `model` in the body, so we need to remove it.
+        /* Azure Bug, extremely short default `max_tokens` response */
+        if (!modelOptions.max_tokens && modelOptions.model === 'gpt-4-vision-preview') {
+          modelOptions.max_tokens = 4000;
+        }
+
+        /* Azure does not accept `model` in the body, so we need to remove it. */
         delete modelOptions.model;
 
         opts.baseURL = this.langchainProxy
@@ -1120,6 +1120,7 @@ ${convo}
       let chatCompletion;
       /** @type {OpenAI} */
       const openai = new OpenAI({
+        fetch: this.fetch,
         apiKey: this.apiKey,
         ...opts,
       });
@@ -1169,8 +1170,10 @@ ${convo}
         });
       }
 
+      const streamRate = this.options.streamRate ?? Constants.DEFAULT_STREAM_RATE;
+
       if (this.message_file_map && this.isOllama) {
-        const ollamaClient = new OllamaClient({ baseURL });
+        const ollamaClient = new OllamaClient({ baseURL, streamRate });
         return await ollamaClient.chatCompletion({
           payload: modelOptions,
           onProgress,
@@ -1179,7 +1182,15 @@ ${convo}
       }
 
       let UnexpectedRoleError = false;
+      /** @type {Promise<void>} */
+      let streamPromise;
+      /** @type {(value: void | PromiseLike<void>) => void} */
+      let streamResolve;
+
       if (modelOptions.stream) {
+        streamPromise = new Promise((resolve) => {
+          streamResolve = resolve;
+        });
         const stream = await openai.beta.chat.completions
           .stream({
             ...modelOptions,
@@ -1191,13 +1202,17 @@ ${convo}
           .on('error', (err) => {
             handleOpenAIErrors(err, errorCallback, 'stream');
           })
-          .on('finalChatCompletion', (finalChatCompletion) => {
+          .on('finalChatCompletion', async (finalChatCompletion) => {
             const finalMessage = finalChatCompletion?.choices?.[0]?.message;
-            if (finalMessage && finalMessage?.role !== 'assistant') {
+            if (!finalMessage) {
+              return;
+            }
+            await streamPromise;
+            if (finalMessage?.role !== 'assistant') {
               finalChatCompletion.choices[0].message.role = 'assistant';
             }
 
-            if (finalMessage && !finalMessage?.content?.trim()) {
+            if (typeof finalMessage.content !== 'string' || finalMessage.content.trim() === '') {
               finalChatCompletion.choices[0].message.content = intermediateReply;
             }
           })
@@ -1208,7 +1223,6 @@ ${convo}
             }
           });
 
-        const azureDelay = this.modelOptions.model?.includes('gpt-4') ? 30 : 17;
         for await (const chunk of stream) {
           const token = chunk.choices[0]?.delta?.content || '';
           intermediateReply += token;
@@ -1218,11 +1232,11 @@ ${convo}
             break;
           }
 
-          if (this.azure) {
-            await sleep(azureDelay);
-          }
+          await sleep(streamRate);
         }
 
+        streamResolve();
+
         if (!UnexpectedRoleError) {
           chatCompletion = await stream.finalChatCompletion().catch((err) => {
             handleOpenAIErrors(err, errorCallback, 'finalChatCompletion');
@@ -1250,14 +1264,23 @@ ${convo}
         throw new Error('Chat completion failed');
       }
 
-      const { message, finish_reason } = chatCompletion.choices[0];
-      if (chatCompletion) {
-        this.metadata = { finish_reason };
+      const { choices } = chatCompletion;
+      if (!Array.isArray(choices) || choices.length === 0) {
+        logger.warn('[OpenAIClient] Chat completion response has no choices');
+        return intermediateReply;
       }
 
+      const { message, finish_reason } = choices[0] ?? {};
+      this.metadata = { finish_reason };
+
       logger.debug('[OpenAIClient] chatCompletion response', chatCompletion);
 
-      if (!message?.content?.trim() && intermediateReply.length) {
+      if (!message) {
+        logger.warn('[OpenAIClient] Message is undefined in chatCompletion response');
+        return intermediateReply;
+      }
+
+      if (typeof message.content !== 'string' || message.content.trim() === '') {
         logger.debug(
           '[OpenAIClient] chatCompletion: using intermediateReply due to empty message.content',
           { intermediateReply },

api/app/clients/PluginsClient.js

@@ -1,5 +1,6 @@
 const OpenAIClient = require('./OpenAIClient');
 const { CallbackManager } = require('langchain/callbacks');
+const { CacheKeys, Time } = require('librechat-data-provider');
 const { BufferMemory, ChatMessageHistory } = require('langchain/memory');
 const { initializeCustomAgent, initializeFunctionsAgent } = require('./agents');
 const { addImages, buildErrorInput, buildPromptPrefix } = require('./output_parsers');
@@ -11,6 +12,7 @@ const { SelfReflectionTool } = require('./tools');
 const { isEnabled } = require('~/server/utils');
 const { extractBaseURL } = require('~/utils');
 const { loadTools } = require('./tools/util');
+const { getLogStores } = require('~/cache');
 const { logger } = require('~/config');
 
 class PluginsClient extends OpenAIClient {
@@ -220,6 +222,13 @@ class PluginsClient extends OpenAIClient {
     }
   }
 
+  /**
+   *
+   * @param {TMessage} responseMessage
+   * @param {Partial<TMessage>} saveOptions
+   * @param {string} user
+   * @returns
+   */
   async handleResponseMessage(responseMessage, saveOptions, user) {
     const { output, errorMessage, ...result } = this.result;
     logger.debug('[PluginsClient][handleResponseMessage] Output:', {
@@ -238,18 +247,39 @@ class PluginsClient extends OpenAIClient {
       await this.recordTokenUsage(responseMessage);
     }
 
-    await this.saveMessageToDatabase(responseMessage, saveOptions, user);
+    this.responsePromise = this.saveMessageToDatabase(responseMessage, saveOptions, user);
+    const messageCache = getLogStores(CacheKeys.MESSAGES);
+    messageCache.set(
+      responseMessage.messageId,
+      {
+        text: responseMessage.text,
+        complete: true,
+      },
+      Time.FIVE_MINUTES,
+    );
     delete responseMessage.tokenCount;
     return { ...responseMessage, ...result };
   }
 
   async sendMessage(message, opts = {}) {
+    /** @type {{ filteredTools: string[], includedTools: string[] }} */
+    const { filteredTools = [], includedTools = [] } = this.options.req.app.locals;
+
+    if (includedTools.length > 0) {
+      const tools = this.options.tools.filter((plugin) => includedTools.includes(plugin));
+      this.options.tools = tools;
+    } else {
+      const tools = this.options.tools.filter((plugin) => !filteredTools.includes(plugin));
+      this.options.tools = tools;
+    }
+
     // If a message is edited, no tools can be used.
     const completionMode = this.options.tools.length === 0 || opts.isEdited;
     if (completionMode) {
       this.setOptions(opts);
       return super.sendMessage(message, opts);
     }
+
     logger.debug('[PluginsClient] sendMessage', { userMessageText: message, opts });
     const {
       user,
@@ -264,6 +294,14 @@ class PluginsClient extends OpenAIClient {
       onToolEnd,
     } = await this.handleStartMethods(message, opts);
 
+    if (opts.progressCallback) {
+      opts.onProgress = opts.progressCallback.call(null, {
+        ...(opts.progressOptions ?? {}),
+        parentMessageId: userMessage.messageId,
+        messageId: responseMessageId,
+      });
+    }
+
     this.currentMessages.push(userMessage);
 
     let {
@@ -292,7 +330,15 @@ class PluginsClient extends OpenAIClient {
     if (payload) {
       this.currentMessages = payload;
     }
-    await this.saveMessageToDatabase(userMessage, saveOptions, user);
+
+    if (!this.skipSaveUserMessage) {
+      this.userMessagePromise = this.saveMessageToDatabase(userMessage, saveOptions, user);
+      if (typeof opts?.getReqData === 'function') {
+        opts.getReqData({
+          userMessagePromise: this.userMessagePromise,
+        });
+      }
+    }
 
     if (isEnabled(process.env.CHECK_BALANCE)) {
       await checkBalance({

api/app/clients/agents/CustomAgent/instructions.js

@@ -1,44 +1,3 @@
-/*
-module.exports = `You are ChatGPT, a Large Language model with useful tools.
-
-Talk to the human and provide meaningful answers when questions are asked.
-
-Use the tools when you need them, but use your own knowledge if you are confident of the answer. Keep answers short and concise.
-
-A tool is not usually needed for creative requests, so do your best to answer them without tools.
-
-Avoid repeating identical answers if it appears before. Only fulfill the human's requests, do not create extra steps beyond what the human has asked for.
-
-Your input for 'Action' should be the name of tool used only.
-
-Be honest. If you can't answer something, or a tool is not appropriate, say you don't know or answer to the best of your ability.
-
-Attempt to fulfill the human's requests in as few actions as possible`;
-*/
-
-// module.exports = `You are ChatGPT, a highly knowledgeable and versatile large language model.
-
-// Engage with the Human conversationally, providing concise and meaningful answers to questions. Utilize built-in tools when necessary, except for creative requests, where relying on your own knowledge is preferred. Aim for variety and avoid repetitive answers.
-
-// For your 'Action' input, state the name of the tool used only, and honor user requests without adding extra steps. Always be honest; if you cannot provide an appropriate answer or tool, admit that or do your best.
-
-// Strive to meet the user's needs efficiently with minimal actions.`;
-
-// import {
-//   BasePromptTemplate,
-//   BaseStringPromptTemplate,
-//   SerializedBasePromptTemplate,
-//   renderTemplate,
-// } from "langchain/prompts";
-
-// prefix: `You are ChatGPT, a highly knowledgeable and versatile large language model.
-// Your objective is to help users by understanding their intent and choosing the best action. Prioritize direct, specific responses. Use concise, varied answers and rely on your knowledge for creative tasks. Utilize tools when needed, and structure results for machine compatibility.
-// prefix: `Objective: to comprehend human intentions based on user input and available tools. Goal: identify the best action to directly address the human's query. In your subsequent steps, you will utilize the chosen action. You may select multiple actions and list them in a meaningful order. Prioritize actions that directly relate to the user's query over general ones. Ensure that the generated thought is highly specific and explicit to best match the user's expectations. Construct the result in a manner that an online open-API would most likely expect. Provide concise and meaningful answers to human queries. Utilize tools when necessary. Relying on your own knowledge is preferred for creative requests. Aim for variety and avoid repetitive answers.
-
-// # Available Actions & Tools:
-// N/A: no suitable action, use your own knowledge.`,
-// suffix: `Remember, all your responses MUST adhere to the described format and only respond if the format is followed. Output exactly with the requested format, avoiding any other text as this will be parsed by a machine. Following 'Action:', provide only one of the actions listed above. If a tool is not necessary, deduce this quickly and finish your response. Honor the human's requests without adding extra steps. Carry out tasks in the sequence written by the human. Always be honest; if you cannot provide an appropriate answer or tool, do your best with your own knowledge. Strive to meet the user's needs efficiently with minimal actions.`;
-
 module.exports = {
   'gpt3-v1': {
     prefix: `Objective: Understand human intentions using user input and available tools. Goal: Identify the most suitable actions to directly address user queries.

api/app/clients/llm/RunManager.js

@@ -1,5 +1,5 @@
 const { createStartHandler } = require('~/app/clients/callbacks');
-const spendTokens = require('~/models/spendTokens');
+const { spendTokens } = require('~/models/spendTokens');
 const { logger } = require('~/config');
 
 class RunManager {

api/app/clients/output_parsers/addImages.js

@@ -60,10 +60,10 @@ function addImages(intermediateSteps, responseMessage) {
     if (!observation || !observation.includes('![')) {
       return;
     }
-    const observedImagePath = observation.match(/!\[.*\]\([^)]*\)/g);
+    const observedImagePath = observation.match(/!\[[^(]*\]\([^)]*\)/g);
     if (observedImagePath && !responseMessage.text.includes(observedImagePath[0])) {
-      responseMessage.text += '\n' + observation;
-      logger.debug('[addImages] added image from intermediateSteps:', observation);
+      responseMessage.text += '\n' + observedImagePath[0];
+      logger.debug('[addImages] added image from intermediateSteps:', observedImagePath[0]);
     }
   });
 }

api/app/clients/output_parsers/addImages.spec.js

@@ -81,4 +81,62 @@ describe('addImages', () => {
     addImages(intermediateSteps, responseMessage);
     expect(responseMessage.text).toBe(`${originalText}\n${imageMarkdown}`);
   });
+
+  it('should extract only image markdowns when there is text between them', () => {
+    const markdownWithTextBetweenImages = `
+      ![image1](/images/image1.png)
+      Some text between images that should not be included.
+      ![image2](/images/image2.png)
+      More text that should be ignored.
+      ![image3](/images/image3.png)
+    `;
+    intermediateSteps.push({ observation: markdownWithTextBetweenImages });
+    addImages(intermediateSteps, responseMessage);
+    expect(responseMessage.text).toBe('\n![image1](/images/image1.png)');
+  });
+
+  it('should only return the first image when multiple images are present', () => {
+    const markdownWithMultipleImages = `
+      ![image1](/images/image1.png)
+      ![image2](/images/image2.png)
+      ![image3](/images/image3.png)
+    `;
+    intermediateSteps.push({ observation: markdownWithMultipleImages });
+    addImages(intermediateSteps, responseMessage);
+    expect(responseMessage.text).toBe('\n![image1](/images/image1.png)');
+  });
+
+  it('should not include any text or metadata surrounding the image markdown', () => {
+    const markdownWithMetadata = `
+      Title: Test Document
+      Author: John Doe
+      ![image1](/images/image1.png)
+      Some content after the image.
+      Vector values: [0.1, 0.2, 0.3]
+    `;
+    intermediateSteps.push({ observation: markdownWithMetadata });
+    addImages(intermediateSteps, responseMessage);
+    expect(responseMessage.text).toBe('\n![image1](/images/image1.png)');
+  });
+
+  it('should handle complex markdown with multiple images and only return the first one', () => {
+    const complexMarkdown = `
+      # Document Title
+      
+      ## Section 1
+      Here's some text with an embedded image:
+      ![image1](/images/image1.png)
+      
+      ## Section 2
+      More text here...
+      ![image2](/images/image2.png)
+      
+      ### Subsection
+      Even more content
+      ![image3](/images/image3.png)
+    `;
+    intermediateSteps.push({ observation: complexMarkdown });
+    addImages(intermediateSteps, responseMessage);
+    expect(responseMessage.text).toBe('\n![image1](/images/image1.png)');
+  });
 });

api/app/clients/prompts/addCacheControl.js

@@ -0,0 +1,43 @@
+/**
+ * Anthropic API: Adds cache control to the appropriate user messages in the payload.
+ * @param {Array<AnthropicMessage>} messages - The array of message objects.
+ * @returns {Array<AnthropicMessage>} - The updated array of message objects with cache control added.
+ */
+function addCacheControl(messages) {
+  if (!Array.isArray(messages) || messages.length < 2) {
+    return messages;
+  }
+
+  const updatedMessages = [...messages];
+  let userMessagesModified = 0;
+
+  for (let i = updatedMessages.length - 1; i >= 0 && userMessagesModified < 2; i--) {
+    const message = updatedMessages[i];
+    if (message.role !== 'user') {
+      continue;
+    }
+
+    if (typeof message.content === 'string') {
+      message.content = [
+        {
+          type: 'text',
+          text: message.content,
+          cache_control: { type: 'ephemeral' },
+        },
+      ];
+      userMessagesModified++;
+    } else if (Array.isArray(message.content)) {
+      for (let j = message.content.length - 1; j >= 0; j--) {
+        if (message.content[j].type === 'text') {
+          message.content[j].cache_control = { type: 'ephemeral' };
+          userMessagesModified++;
+          break;
+        }
+      }
+    }
+  }
+
+  return updatedMessages;
+}
+
+module.exports = addCacheControl;

api/app/clients/prompts/addCacheControl.spec.js

@@ -0,0 +1,227 @@
+const addCacheControl = require('./addCacheControl');
+
+describe('addCacheControl', () => {
+  test('should add cache control to the last two user messages with array content', () => {
+    const messages = [
+      { role: 'user', content: [{ type: 'text', text: 'Hello' }] },
+      { role: 'assistant', content: [{ type: 'text', text: 'Hi there' }] },
+      { role: 'user', content: [{ type: 'text', text: 'How are you?' }] },
+      { role: 'assistant', content: [{ type: 'text', text: 'I\'m doing well, thanks!' }] },
+      { role: 'user', content: [{ type: 'text', text: 'Great!' }] },
+    ];
+
+    const result = addCacheControl(messages);
+
+    expect(result[0].content[0]).not.toHaveProperty('cache_control');
+    expect(result[2].content[0].cache_control).toEqual({ type: 'ephemeral' });
+    expect(result[4].content[0].cache_control).toEqual({ type: 'ephemeral' });
+  });
+
+  test('should add cache control to the last two user messages with string content', () => {
+    const messages = [
+      { role: 'user', content: 'Hello' },
+      { role: 'assistant', content: 'Hi there' },
+      { role: 'user', content: 'How are you?' },
+      { role: 'assistant', content: 'I\'m doing well, thanks!' },
+      { role: 'user', content: 'Great!' },
+    ];
+
+    const result = addCacheControl(messages);
+
+    expect(result[0].content).toBe('Hello');
+    expect(result[2].content[0]).toEqual({
+      type: 'text',
+      text: 'How are you?',
+      cache_control: { type: 'ephemeral' },
+    });
+    expect(result[4].content[0]).toEqual({
+      type: 'text',
+      text: 'Great!',
+      cache_control: { type: 'ephemeral' },
+    });
+  });
+
+  test('should handle mixed string and array content', () => {
+    const messages = [
+      { role: 'user', content: 'Hello' },
+      { role: 'assistant', content: 'Hi there' },
+      { role: 'user', content: [{ type: 'text', text: 'How are you?' }] },
+    ];
+
+    const result = addCacheControl(messages);
+
+    expect(result[0].content[0]).toEqual({
+      type: 'text',
+      text: 'Hello',
+      cache_control: { type: 'ephemeral' },
+    });
+    expect(result[2].content[0].cache_control).toEqual({ type: 'ephemeral' });
+  });
+
+  test('should handle less than two user messages', () => {
+    const messages = [
+      { role: 'user', content: 'Hello' },
+      { role: 'assistant', content: 'Hi there' },
+    ];
+
+    const result = addCacheControl(messages);
+
+    expect(result[0].content[0]).toEqual({
+      type: 'text',
+      text: 'Hello',
+      cache_control: { type: 'ephemeral' },
+    });
+    expect(result[1].content).toBe('Hi there');
+  });
+
+  test('should return original array if no user messages', () => {
+    const messages = [
+      { role: 'assistant', content: 'Hi there' },
+      { role: 'assistant', content: 'How can I help?' },
+    ];
+
+    const result = addCacheControl(messages);
+
+    expect(result).toEqual(messages);
+  });
+
+  test('should handle empty array', () => {
+    const messages = [];
+    const result = addCacheControl(messages);
+    expect(result).toEqual([]);
+  });
+
+  test('should handle non-array input', () => {
+    const messages = 'not an array';
+    const result = addCacheControl(messages);
+    expect(result).toBe('not an array');
+  });
+
+  test('should not modify assistant messages', () => {
+    const messages = [
+      { role: 'user', content: 'Hello' },
+      { role: 'assistant', content: 'Hi there' },
+      { role: 'user', content: 'How are you?' },
+    ];
+
+    const result = addCacheControl(messages);
+
+    expect(result[1].content).toBe('Hi there');
+  });
+
+  test('should handle multiple content items in user messages', () => {
+    const messages = [
+      {
+        role: 'user',
+        content: [
+          { type: 'text', text: 'Hello' },
+          { type: 'image', url: 'http://example.com/image.jpg' },
+          { type: 'text', text: 'This is an image' },
+        ],
+      },
+      { role: 'assistant', content: 'Hi there' },
+      { role: 'user', content: 'How are you?' },
+    ];
+
+    const result = addCacheControl(messages);
+
+    expect(result[0].content[0]).not.toHaveProperty('cache_control');
+    expect(result[0].content[1]).not.toHaveProperty('cache_control');
+    expect(result[0].content[2].cache_control).toEqual({ type: 'ephemeral' });
+    expect(result[2].content[0]).toEqual({
+      type: 'text',
+      text: 'How are you?',
+      cache_control: { type: 'ephemeral' },
+    });
+  });
+
+  test('should handle an array with mixed content types', () => {
+    const messages = [
+      { role: 'user', content: 'Hello' },
+      { role: 'assistant', content: 'Hi there' },
+      { role: 'user', content: [{ type: 'text', text: 'How are you?' }] },
+      { role: 'assistant', content: 'I\'m doing well, thanks!' },
+      { role: 'user', content: 'Great!' },
+    ];
+
+    const result = addCacheControl(messages);
+
+    expect(result[0].content).toEqual('Hello');
+    expect(result[2].content[0]).toEqual({
+      type: 'text',
+      text: 'How are you?',
+      cache_control: { type: 'ephemeral' },
+    });
+    expect(result[4].content).toEqual([
+      {
+        type: 'text',
+        text: 'Great!',
+        cache_control: { type: 'ephemeral' },
+      },
+    ]);
+    expect(result[1].content).toBe('Hi there');
+    expect(result[3].content).toBe('I\'m doing well, thanks!');
+  });
+
+  test('should handle edge case with multiple content types', () => {
+    const messages = [
+      {
+        role: 'user',
+        content: [
+          {
+            type: 'image',
+            source: { type: 'base64', media_type: 'image/png', data: 'some_base64_string' },
+          },
+          {
+            type: 'image',
+            source: { type: 'base64', media_type: 'image/png', data: 'another_base64_string' },
+          },
+          { type: 'text', text: 'what do all these images have in common' },
+        ],
+      },
+      { role: 'assistant', content: 'I see multiple images.' },
+      { role: 'user', content: 'Correct!' },
+    ];
+
+    const result = addCacheControl(messages);
+
+    expect(result[0].content[0]).not.toHaveProperty('cache_control');
+    expect(result[0].content[1]).not.toHaveProperty('cache_control');
+    expect(result[0].content[2].cache_control).toEqual({ type: 'ephemeral' });
+    expect(result[2].content[0]).toEqual({
+      type: 'text',
+      text: 'Correct!',
+      cache_control: { type: 'ephemeral' },
+    });
+  });
+
+  test('should handle user message with no text block', () => {
+    const messages = [
+      {
+        role: 'user',
+        content: [
+          {
+            type: 'image',
+            source: { type: 'base64', media_type: 'image/png', data: 'some_base64_string' },
+          },
+          {
+            type: 'image',
+            source: { type: 'base64', media_type: 'image/png', data: 'another_base64_string' },
+          },
+        ],
+      },
+      { role: 'assistant', content: 'I see two images.' },
+      { role: 'user', content: 'Correct!' },
+    ];
+
+    const result = addCacheControl(messages);
+
+    expect(result[0].content[0]).not.toHaveProperty('cache_control');
+    expect(result[0].content[1]).not.toHaveProperty('cache_control');
+    expect(result[2].content[0]).toEqual({
+      type: 'text',
+      text: 'Correct!',
+      cache_control: { type: 'ephemeral' },
+    });
+  });
+});

api/app/clients/prompts/artifacts.js

@@ -0,0 +1,162 @@
+const artifactsPrompt = `The assistant can create and reference artifacts during conversations.
+  
+Artifacts are for substantial, self-contained content that users might modify or reuse, displayed in a separate UI window for clarity.
+
+# Good artifacts are...
+- Substantial content (>15 lines)
+- Content that the user is likely to modify, iterate on, or take ownership of
+- Self-contained, complex content that can be understood on its own, without context from the conversation
+- Content intended for eventual use outside the conversation (e.g., reports, emails, presentations)
+- Content likely to be referenced or reused multiple times
+
+# Don't use artifacts for...
+- Simple, informational, or short content, such as brief code snippets, mathematical equations, or small examples
+- Primarily explanatory, instructional, or illustrative content, such as examples provided to clarify a concept
+- Suggestions, commentary, or feedback on existing artifacts
+- Conversational or explanatory content that doesn't represent a standalone piece of work
+- Content that is dependent on the current conversational context to be useful
+- Content that is unlikely to be modified or iterated upon by the user
+- Request from users that appears to be a one-off question
+
+# Usage notes
+- One artifact per message unless specifically requested
+- Prefer in-line content (don't use artifacts) when possible. Unnecessary use of artifacts can be jarring for users.
+- If a user asks the assistant to "draw an SVG" or "make a website," the assistant does not need to explain that it doesn't have these capabilities. Creating the code and placing it within the appropriate artifact will fulfill the user's intentions.
+- If asked to generate an image, the assistant can offer an SVG instead. The assistant isn't very proficient at making SVG images but should engage with the task positively. Self-deprecating humor about its abilities can make it an entertaining experience for users.
+- The assistant errs on the side of simplicity and avoids overusing artifacts for content that can be effectively presented within the conversation.
+- Always provide complete, specific, and fully functional content without any placeholders, ellipses, or 'remains the same' comments.
+
+<artifact_instructions>
+  When collaborating with the user on creating content that falls into compatible categories, the assistant should follow these steps:
+
+  1. Create the artifact using the following format:
+
+     :::artifact{identifier="unique-identifier" type="mime-type" title="Artifact Title"}
+     \`\`\`
+     Your artifact content here
+     \`\`\`
+     :::
+
+  2. Assign an identifier to the \`identifier\` attribute. For updates, reuse the prior identifier. For new artifacts, the identifier should be descriptive and relevant to the content, using kebab-case (e.g., "example-code-snippet"). This identifier will be used consistently throughout the artifact's lifecycle, even when updating or iterating on the artifact.
+  3. Include a \`title\` attribute to provide a brief title or description of the content.
+  4. Add a \`type\` attribute to specify the type of content the artifact represents. Assign one of the following values to the \`type\` attribute:
+    - Code: "application/vnd.code"
+      - Use for code snippets or scripts in any programming language.
+      - Include the language name as the value of the \`language\` attribute (e.g., \`language="python"\`).
+    - Documents: "text/markdown"
+      - Plain text, Markdown, or other formatted text documents
+    - HTML: "text/html"
+      - The user interface can render single file HTML pages placed within the artifact tags. HTML, JS, and CSS should be in a single file when using the \`text/html\` type.
+      - Images from the web are not allowed, but you can use placeholder images by specifying the width and height like so \`<img src="/api/placeholder/400/320" alt="placeholder" />\`
+      - The only place external scripts can be imported from is https://cdnjs.cloudflare.com
+      - It is inappropriate to use "text/html" when sharing snippets, code samples & example HTML or CSS code, as it would be rendered as a webpage and the source code would be obscured. The assistant should instead use "application/vnd.code" defined above.
+      - If the assistant is unable to follow the above requirements for any reason, use "application/vnd.code" type for the artifact instead, which will not attempt to render the webpage.
+    - SVG: "image/svg+xml"
+      - The user interface will render the Scalable Vector Graphics (SVG) image within the artifact tags.
+      - The assistant should specify the viewbox of the SVG rather than defining a width/height
+    - Mermaid Diagrams: "application/vnd.mermaid"
+      - The user interface will render Mermaid diagrams placed within the artifact tags.
+    - React Components: "application/vnd.react"
+      - Use this for displaying either: React elements, e.g. \`<strong>Hello World!</strong>\`, React pure functional components, e.g. \`() => <strong>Hello World!</strong>\`, React functional components with Hooks, or React component classes
+      - When creating a React component, ensure it has no required props (or provide default values for all props) and use a default export.
+      - Use Tailwind classes for styling. DO NOT USE ARBITRARY VALUES (e.g. \`h-[600px]\`).
+      - Base React is available to be imported. To use hooks, first import it at the top of the artifact, e.g. \`import { useState } from "react"\`
+      - The lucide-react@0.263.1 library is available to be imported. e.g. \`import { Camera } from "lucide-react"\` & \`<Camera color="red" size={48} />\`
+      - The recharts charting library is available to be imported, e.g. \`import { LineChart, XAxis, ... } from "recharts"\` & \`<LineChart ...><XAxis dataKey="name"> ...\`
+      - The assistant can use prebuilt components from the \`shadcn/ui\` library after it is imported: \`import { Alert, AlertDescription, AlertTitle, AlertDialog, AlertDialogAction } from '/components/ui/alert';\`. If using components from the shadcn/ui library, the assistant mentions this to the user and offers to help them install the components if necessary.
+      - Components MUST be imported from \`/components/ui/name\` and NOT from \`/components/name\` or \`@/components/ui/name\`.
+      - NO OTHER LIBRARIES (e.g. zod, hookform) ARE INSTALLED OR ABLE TO BE IMPORTED.
+      - Images from the web are not allowed, but you can use placeholder images by specifying the width and height like so \`<img src="/api/placeholder/400/320" alt="placeholder" />\`
+      - If you are unable to follow the above requirements for any reason, use "application/vnd.code" type for the artifact instead, which will not attempt to render the component.
+  5. Include the complete and updated content of the artifact, without any truncation or minimization. Don't use "// rest of the code remains the same...".
+  6. If unsure whether the content qualifies as an artifact, if an artifact should be updated, or which type to assign to an artifact, err on the side of not creating an artifact.
+  7. Always use triple backticks (\`\`\`) to enclose the content within the artifact, regardless of the content type.
+</artifact_instructions>
+
+Here are some examples of correct usage of artifacts:
+
+<examples>
+  <example_docstring>
+    This example demonstrates how to create a new artifact for a Python script.
+  </example_docstring>
+
+  <example>
+    <user_query>Can you help me create a Python script to calculate the factorial of a number?</user_query>
+
+    <assistant_response>
+      Sure! Here's a Python script that calculates the factorial of a number:
+
+      :::artifact{identifier="factorial-script" type="application/vnd.code" language="python" title="Simple Python factorial script"}
+      \`\`\`python
+      def factorial(n):
+          if n == 0:
+              return 1
+          else:
+              return n * factorial(n - 1)
+
+      # Get user input
+      num = int(input("Enter a non-negative integer: "))
+
+      # Calculate and print the factorial
+      result = factorial(num)
+      print(f"The factorial of {num} is {result}")
+      \`\`\`
+      :::
+
+      This script defines a \`factorial\` function using recursion to calculate the factorial of a given number. It then prompts the user for input, calculates the factorial, and prints the result.
+
+      To use this script:
+      1. Copy the code into a .py file
+      2. Run the script
+      3. Enter a non-negative integer when prompted
+      4. The script will output the factorial of the number you entered
+
+      Let me know if you have any questions or if you'd like to modify the script in any way!
+    </assistant_response>
+  </example>
+
+  <example_docstring>
+    This example demonstrates how to create a Mermaid artifact for a simple flow chart.
+  </example_docstring>
+
+  <example>
+    <user_query>Can you create a simple flow chart showing the process of making tea using Mermaid?</user_query>
+
+    <assistant_response>
+      Sure! Here's a simple flow chart depicting the process of making tea using Mermaid syntax:
+
+      :::artifact{identifier="tea-making-flowchart" type="application/vnd.mermaid" title="Flow chart: Making Tea"}
+      \`\`\`mermaid
+      graph TD
+          A[Start] --> B{Water boiled?}
+          B -->|Yes| C[Add tea leaves to cup]
+          B -->|No| D[Boil water]
+          D --> B
+          C --> E[Pour boiling water into cup]
+          E --> F[Steep tea for desired time]
+          F --> G[Remove tea leaves]
+          G --> H[Add milk or sugar, if desired]
+          H --> I[Enjoy your tea!]
+          I --> J[End]
+      \`\`\`
+      :::
+
+      This flow chart uses Mermaid syntax to visualize the steps involved in making a cup of tea. Here's a brief explanation of the process:
+
+      1. Start
+      2. Check if water is boiled
+      3. If not boiled, boil the water
+      4. Once water is boiled, add tea leaves to the cup
+      5. Pour boiling water into the cup
+      6. Steep the tea for the desired time
+      7. Remove the tea leaves
+      8. Optionally add milk or sugar
+      9. Enjoy your tea!
+      10. End
+
+      This chart provides a clear visual representation of the tea-making process. You can easily modify or expand this chart if you want to add more details or steps to the process. Let me know if you'd like any changes or have any questions!
+    </assistant_response>
+  </example>
+</examples>`;
+
+module.exports = artifactsPrompt;

api/app/clients/prompts/createContextHandlers.js

@@ -8,8 +8,6 @@ In your response, remember to follow these guidelines:
 - If you don't know the answer, simply say that you don't know.
 - If you are unsure how to answer, ask for clarification.
 - Avoid mentioning that you obtained the information from the context.
-
-Answer appropriately in the user's language.
 `;
 
 function createContextHandlers(req, userMessageContent) {
@@ -94,37 +92,40 @@ function createContextHandlers(req, userMessageContent) {
 
       const resolvedQueries = await Promise.all(queryPromises);
 
-      const context = resolvedQueries
-        .map((queryResult, index) => {
-          const file = processedFiles[index];
-          let contextItems = queryResult.data;
+      const context =
+        resolvedQueries.length === 0
+          ? '\n\tThe semantic search did not return any results.'
+          : resolvedQueries
+            .map((queryResult, index) => {
+              const file = processedFiles[index];
+              let contextItems = queryResult.data;
 
-          const generateContext = (currentContext) =>
-            `
+              const generateContext = (currentContext) =>
+                `
           <file>
             <filename>${file.filename}</filename>
             <context>${currentContext}
             </context>
           </file>`;
 
-          if (useFullContext) {
-            return generateContext(`\n${contextItems}`);
-          }
+              if (useFullContext) {
+                return generateContext(`\n${contextItems}`);
+              }
 
-          contextItems = queryResult.data
-            .map((item) => {
-              const pageContent = item[0].page_content;
-              return `
+              contextItems = queryResult.data
+                .map((item) => {
+                  const pageContent = item[0].page_content;
+                  return `
             <contextItem>
               <![CDATA[${pageContent?.trim()}]]>
             </contextItem>`;
+                })
+                .join('');
+
+              return generateContext(contextItems);
             })
             .join('');
 
-          return generateContext(contextItems);
-        })
-        .join('');
-
       if (useFullContext) {
         const prompt = `${header}
           ${context}

api/app/clients/prompts/index.js

@@ -1,3 +1,4 @@
+const addCacheControl = require('./addCacheControl');
 const formatMessages = require('./formatMessages');
 const summaryPrompts = require('./summaryPrompts');
 const handleInputs = require('./handleInputs');
@@ -8,6 +9,7 @@ const createVisionPrompt = require('./createVisionPrompt');
 const createContextHandlers = require('./createContextHandlers');
 
 module.exports = {
+  addCacheControl,
   ...formatMessages,
   ...summaryPrompts,
   ...handleInputs,

api/app/clients/prompts/titlePrompts.js

@@ -28,7 +28,7 @@ ${convo}`,
 };
 
 const titleInstruction =
-  'a concise, 5-word-or-less title for the conversation, using its same language, with no punctuation. Apply title case conventions appropriate for the language. For English, use AP Stylebook Title Case. Never directly mention the language name or the word "title"';
+  'a concise, 5-word-or-less title for the conversation, using its same language, with no punctuation. Apply title case conventions appropriate for the language. Never directly mention the language name or the word "title"';
 const titleFunctionPrompt = `In this environment you have access to a set of tools you can use to generate the conversation title.
   
 You may call them like this:

api/app/clients/specs/AnthropicClient.test.js

@@ -1,4 +1,6 @@
-const AnthropicClient = require('../AnthropicClient');
+const { anthropicSettings } = require('librechat-data-provider');
+const AnthropicClient = require('~/app/clients/AnthropicClient');
+
 const HUMAN_PROMPT = '\n\nHuman:';
 const AI_PROMPT = '\n\nAssistant:';
 
@@ -22,7 +24,7 @@ describe('AnthropicClient', () => {
     const options = {
       modelOptions: {
         model,
-        temperature: 0.7,
+        temperature: anthropicSettings.temperature.default,
       },
     };
     client = new AnthropicClient('test-api-key');
@@ -33,7 +35,42 @@ describe('AnthropicClient', () => {
     it('should set the options correctly', () => {
       expect(client.apiKey).toBe('test-api-key');
       expect(client.modelOptions.model).toBe(model);
-      expect(client.modelOptions.temperature).toBe(0.7);
+      expect(client.modelOptions.temperature).toBe(anthropicSettings.temperature.default);
+    });
+
+    it('should set legacy maxOutputTokens for non-Claude-3 models', () => {
+      const client = new AnthropicClient('test-api-key');
+      client.setOptions({
+        modelOptions: {
+          model: 'claude-2',
+          maxOutputTokens: anthropicSettings.maxOutputTokens.default,
+        },
+      });
+      expect(client.modelOptions.maxOutputTokens).toBe(
+        anthropicSettings.legacy.maxOutputTokens.default,
+      );
+    });
+    it('should not set maxOutputTokens if not provided', () => {
+      const client = new AnthropicClient('test-api-key');
+      client.setOptions({
+        modelOptions: {
+          model: 'claude-3',
+        },
+      });
+      expect(client.modelOptions.maxOutputTokens).toBeUndefined();
+    });
+
+    it('should not set legacy maxOutputTokens for Claude-3 models', () => {
+      const client = new AnthropicClient('test-api-key');
+      client.setOptions({
+        modelOptions: {
+          model: 'claude-3-opus-20240229',
+          maxOutputTokens: anthropicSettings.legacy.maxOutputTokens.default,
+        },
+      });
+      expect(client.modelOptions.maxOutputTokens).toBe(
+        anthropicSettings.legacy.maxOutputTokens.default,
+      );
     });
   });
 
@@ -136,4 +173,212 @@ describe('AnthropicClient', () => {
       expect(prompt).toContain('You are Claude-2');
     });
   });
+
+  describe('getClient', () => {
+    it('should set legacy maxOutputTokens for non-Claude-3 models', () => {
+      const client = new AnthropicClient('test-api-key');
+      client.setOptions({
+        modelOptions: {
+          model: 'claude-2',
+          maxOutputTokens: anthropicSettings.legacy.maxOutputTokens.default,
+        },
+      });
+      expect(client.modelOptions.maxOutputTokens).toBe(
+        anthropicSettings.legacy.maxOutputTokens.default,
+      );
+    });
+
+    it('should not set legacy maxOutputTokens for Claude-3 models', () => {
+      const client = new AnthropicClient('test-api-key');
+      client.setOptions({
+        modelOptions: {
+          model: 'claude-3-opus-20240229',
+          maxOutputTokens: anthropicSettings.legacy.maxOutputTokens.default,
+        },
+      });
+      expect(client.modelOptions.maxOutputTokens).toBe(
+        anthropicSettings.legacy.maxOutputTokens.default,
+      );
+    });
+
+    it('should add beta header for claude-3-5-sonnet model', () => {
+      const client = new AnthropicClient('test-api-key');
+      const modelOptions = {
+        model: 'claude-3-5-sonnet-20240307',
+      };
+      client.setOptions({ modelOptions, promptCache: true });
+      const anthropicClient = client.getClient(modelOptions);
+      expect(anthropicClient._options.defaultHeaders).toBeDefined();
+      expect(anthropicClient._options.defaultHeaders).toHaveProperty('anthropic-beta');
+      expect(anthropicClient._options.defaultHeaders['anthropic-beta']).toBe(
+        'max-tokens-3-5-sonnet-2024-07-15,prompt-caching-2024-07-31',
+      );
+    });
+
+    it('should add beta header for claude-3-haiku model', () => {
+      const client = new AnthropicClient('test-api-key');
+      const modelOptions = {
+        model: 'claude-3-haiku-2028',
+      };
+      client.setOptions({ modelOptions, promptCache: true });
+      const anthropicClient = client.getClient(modelOptions);
+      expect(anthropicClient._options.defaultHeaders).toBeDefined();
+      expect(anthropicClient._options.defaultHeaders).toHaveProperty('anthropic-beta');
+      expect(anthropicClient._options.defaultHeaders['anthropic-beta']).toBe(
+        'prompt-caching-2024-07-31',
+      );
+    });
+
+    it('should not add beta header for other models', () => {
+      const client = new AnthropicClient('test-api-key');
+      client.setOptions({
+        modelOptions: {
+          model: 'claude-2',
+        },
+      });
+      const anthropicClient = client.getClient();
+      expect(anthropicClient.defaultHeaders).not.toHaveProperty('anthropic-beta');
+    });
+  });
+
+  describe('calculateCurrentTokenCount', () => {
+    let client;
+
+    beforeEach(() => {
+      client = new AnthropicClient('test-api-key');
+    });
+
+    it('should calculate correct token count when usage is provided', () => {
+      const tokenCountMap = {
+        msg1: 10,
+        msg2: 20,
+        currentMsg: 30,
+      };
+      const currentMessageId = 'currentMsg';
+      const usage = {
+        input_tokens: 70,
+        output_tokens: 50,
+      };
+
+      const result = client.calculateCurrentTokenCount({ tokenCountMap, currentMessageId, usage });
+
+      expect(result).toBe(40); // 70 - (10 + 20) = 40
+    });
+
+    it('should return original estimate if calculation results in negative value', () => {
+      const tokenCountMap = {
+        msg1: 40,
+        msg2: 50,
+        currentMsg: 30,
+      };
+      const currentMessageId = 'currentMsg';
+      const usage = {
+        input_tokens: 80,
+        output_tokens: 50,
+      };
+
+      const result = client.calculateCurrentTokenCount({ tokenCountMap, currentMessageId, usage });
+
+      expect(result).toBe(30); // Original estimate
+    });
+
+    it('should handle cache creation and read input tokens', () => {
+      const tokenCountMap = {
+        msg1: 10,
+        msg2: 20,
+        currentMsg: 30,
+      };
+      const currentMessageId = 'currentMsg';
+      const usage = {
+        input_tokens: 50,
+        cache_creation_input_tokens: 10,
+        cache_read_input_tokens: 20,
+        output_tokens: 40,
+      };
+
+      const result = client.calculateCurrentTokenCount({ tokenCountMap, currentMessageId, usage });
+
+      expect(result).toBe(50); // (50 + 10 + 20) - (10 + 20) = 50
+    });
+
+    it('should handle missing usage properties', () => {
+      const tokenCountMap = {
+        msg1: 10,
+        msg2: 20,
+        currentMsg: 30,
+      };
+      const currentMessageId = 'currentMsg';
+      const usage = {
+        output_tokens: 40,
+      };
+
+      const result = client.calculateCurrentTokenCount({ tokenCountMap, currentMessageId, usage });
+
+      expect(result).toBe(30); // Original estimate
+    });
+
+    it('should handle empty tokenCountMap', () => {
+      const tokenCountMap = {};
+      const currentMessageId = 'currentMsg';
+      const usage = {
+        input_tokens: 50,
+        output_tokens: 40,
+      };
+
+      const result = client.calculateCurrentTokenCount({ tokenCountMap, currentMessageId, usage });
+
+      expect(result).toBe(50);
+      expect(Number.isNaN(result)).toBe(false);
+    });
+
+    it('should handle zero values in usage', () => {
+      const tokenCountMap = {
+        msg1: 10,
+        currentMsg: 20,
+      };
+      const currentMessageId = 'currentMsg';
+      const usage = {
+        input_tokens: 0,
+        cache_creation_input_tokens: 0,
+        cache_read_input_tokens: 0,
+        output_tokens: 0,
+      };
+
+      const result = client.calculateCurrentTokenCount({ tokenCountMap, currentMessageId, usage });
+
+      expect(result).toBe(20); // Should return original estimate
+      expect(Number.isNaN(result)).toBe(false);
+    });
+
+    it('should handle undefined usage', () => {
+      const tokenCountMap = {
+        msg1: 10,
+        currentMsg: 20,
+      };
+      const currentMessageId = 'currentMsg';
+      const usage = undefined;
+
+      const result = client.calculateCurrentTokenCount({ tokenCountMap, currentMessageId, usage });
+
+      expect(result).toBe(20); // Should return original estimate
+      expect(Number.isNaN(result)).toBe(false);
+    });
+
+    it('should handle non-numeric values in tokenCountMap', () => {
+      const tokenCountMap = {
+        msg1: 'ten',
+        currentMsg: 20,
+      };
+      const currentMessageId = 'currentMsg';
+      const usage = {
+        input_tokens: 30,
+        output_tokens: 10,
+      };
+
+      const result = client.calculateCurrentTokenCount({ tokenCountMap, currentMessageId, usage });
+
+      expect(result).toBe(30); // Should return 30 (input_tokens) - 0 (ignored 'ten') = 30
+      expect(Number.isNaN(result)).toBe(false);
+    });
+  });
 });

api/app/clients/specs/BaseClient.test.js

@@ -1,7 +1,7 @@
 const { Constants } = require('librechat-data-provider');
 const { initializeFakeClient } = require('./FakeClient');
 
-jest.mock('../../../lib/db/connectDb');
+jest.mock('~/lib/db/connectDb');
 jest.mock('~/models', () => ({
   User: jest.fn(),
   Key: jest.fn(),
@@ -576,7 +576,11 @@ describe('BaseClient', () => {
       const onStart = jest.fn();
       const opts = { onStart };
       await TestClient.sendMessage('Hello, world!', opts);
-      expect(onStart).toHaveBeenCalledWith(expect.objectContaining({ text: 'Hello, world!' }));
+
+      expect(onStart).toHaveBeenCalledWith(
+        expect.objectContaining({ text: 'Hello, world!' }),
+        expect.any(String),
+      );
     });
 
     test('saveMessageToDatabase is called with the correct arguments', async () => {
@@ -627,5 +631,32 @@ describe('BaseClient', () => {
         }),
       );
     });
+
+    test('userMessagePromise is awaited before saving response message', async () => {
+      // Mock the saveMessageToDatabase method
+      TestClient.saveMessageToDatabase = jest.fn().mockImplementation(() => {
+        return new Promise((resolve) => setTimeout(resolve, 100)); // Simulate a delay
+      });
+
+      // Send a message
+      const messagePromise = TestClient.sendMessage('Hello, world!');
+
+      // Wait a short time to ensure the user message save has started
+      await new Promise((resolve) => setTimeout(resolve, 50));
+
+      // Check that saveMessageToDatabase has been called once (for the user message)
+      expect(TestClient.saveMessageToDatabase).toHaveBeenCalledTimes(1);
+
+      // Wait for the message to be fully processed
+      await messagePromise;
+
+      // Check that saveMessageToDatabase has been called twice (once for user message, once for response)
+      expect(TestClient.saveMessageToDatabase).toHaveBeenCalledTimes(2);
+
+      // Check the order of calls
+      const calls = TestClient.saveMessageToDatabase.mock.calls;
+      expect(calls[0][0].isCreatedByUser).toBe(true); // First call should be for user message
+      expect(calls[1][0].isCreatedByUser).toBe(false); // Second call should be for response message
+    });
   });
 });

api/app/clients/specs/OpenAIClient.test.js

@@ -144,6 +144,7 @@ describe('OpenAIClient', () => {
 
   const defaultOptions = {
     // debug: true,
+    req: {},
     openaiApiKey: 'new-api-key',
     modelOptions: {
       model,

api/app/clients/specs/OpenAIClient.tokens.js

@@ -38,7 +38,12 @@ const run = async () => {
   "On the other hand, we denounce with righteous indignation and dislike men who are so beguiled and demoralized by the charms of pleasure of the moment, so blinded by desire, that they cannot foresee the pain and trouble that are bound to ensue; and equal blame belongs to those who fail in their duty through weakness of will, which is the same as saying through shrinking from toil and pain. These cases are perfectly simple and easy to distinguish. In a free hour, when our power of choice is untrammelled and when nothing prevents our being able to do what we like best, every pleasure is to be welcomed and every pain avoided. But in certain circumstances and owing to the claims of duty or the obligations of business it will frequently occur that pleasures have to be repudiated and annoyances accepted. The wise man therefore always holds in these matters to this principle of selection: he rejects pleasures to secure other greater pleasures, or else he endures pains to avoid worse pains."
   `;
   const model = 'gpt-3.5-turbo';
-  const maxContextTokens = model === 'gpt-4' ? 8191 : model === 'gpt-4-32k' ? 32767 : 4095; // 1 less than maximum
+  let maxContextTokens = 4095;
+  if (model === 'gpt-4') {
+    maxContextTokens = 8191;
+  } else if (model === 'gpt-4-32k') {
+    maxContextTokens = 32767;
+  }
   const clientOptions = {
     reverseProxyUrl: process.env.OPENAI_REVERSE_PROXY || null,
     maxContextTokens,

api/app/clients/specs/PluginsClient.test.js

@@ -194,6 +194,7 @@ describe('PluginsClient', () => {
       expect(client.getFunctionModelName('')).toBe('gpt-3.5-turbo');
     });
   });
+
   describe('Azure OpenAI tests specific to Plugins', () => {
     // TODO: add more tests for Azure OpenAI integration with Plugins
     // let client;
@@ -220,4 +221,94 @@ describe('PluginsClient', () => {
       spy.mockRestore();
     });
   });
+
+  describe('sendMessage with filtered tools', () => {
+    let TestAgent;
+    const apiKey = 'fake-api-key';
+    const mockTools = [{ name: 'tool1' }, { name: 'tool2' }, { name: 'tool3' }, { name: 'tool4' }];
+
+    beforeEach(() => {
+      TestAgent = new PluginsClient(apiKey, {
+        tools: mockTools,
+        modelOptions: {
+          model: 'gpt-3.5-turbo',
+          temperature: 0,
+          max_tokens: 2,
+        },
+        agentOptions: {
+          model: 'gpt-3.5-turbo',
+        },
+      });
+
+      TestAgent.options.req = {
+        app: {
+          locals: {},
+        },
+      };
+
+      TestAgent.sendMessage = jest.fn().mockImplementation(async () => {
+        const { filteredTools = [], includedTools = [] } = TestAgent.options.req.app.locals;
+
+        if (includedTools.length > 0) {
+          const tools = TestAgent.options.tools.filter((plugin) =>
+            includedTools.includes(plugin.name),
+          );
+          TestAgent.options.tools = tools;
+        } else {
+          const tools = TestAgent.options.tools.filter(
+            (plugin) => !filteredTools.includes(plugin.name),
+          );
+          TestAgent.options.tools = tools;
+        }
+
+        return {
+          text: 'Mocked response',
+          tools: TestAgent.options.tools,
+        };
+      });
+    });
+
+    test('should filter out tools when filteredTools is provided', async () => {
+      TestAgent.options.req.app.locals.filteredTools = ['tool1', 'tool3'];
+      const response = await TestAgent.sendMessage('Test message');
+      expect(response.tools).toHaveLength(2);
+      expect(response.tools).toEqual(
+        expect.arrayContaining([
+          expect.objectContaining({ name: 'tool2' }),
+          expect.objectContaining({ name: 'tool4' }),
+        ]),
+      );
+    });
+
+    test('should only include specified tools when includedTools is provided', async () => {
+      TestAgent.options.req.app.locals.includedTools = ['tool2', 'tool4'];
+      const response = await TestAgent.sendMessage('Test message');
+      expect(response.tools).toHaveLength(2);
+      expect(response.tools).toEqual(
+        expect.arrayContaining([
+          expect.objectContaining({ name: 'tool2' }),
+          expect.objectContaining({ name: 'tool4' }),
+        ]),
+      );
+    });
+
+    test('should prioritize includedTools over filteredTools', async () => {
+      TestAgent.options.req.app.locals.filteredTools = ['tool1', 'tool3'];
+      TestAgent.options.req.app.locals.includedTools = ['tool1', 'tool2'];
+      const response = await TestAgent.sendMessage('Test message');
+      expect(response.tools).toHaveLength(2);
+      expect(response.tools).toEqual(
+        expect.arrayContaining([
+          expect.objectContaining({ name: 'tool1' }),
+          expect.objectContaining({ name: 'tool2' }),
+        ]),
+      );
+    });
+
+    test('should not modify tools when no filters are provided', async () => {
+      const response = await TestAgent.sendMessage('Test message');
+      expect(response.tools).toHaveLength(4);
+      expect(response.tools).toEqual(expect.arrayContaining(mockTools));
+    });
+  });
 });

api/app/clients/tools/structured/GoogleSearch.js

@@ -12,9 +12,15 @@ class GoogleSearchResults extends Tool {
     this.envVarApiKey = 'GOOGLE_SEARCH_API_KEY';
     this.envVarSearchEngineId = 'GOOGLE_CSE_ID';
     this.override = fields.override ?? false;
-    this.apiKey = fields.apiKey ?? getEnvironmentVariable(this.envVarApiKey);
+    this.apiKey = fields[this.envVarApiKey] ?? getEnvironmentVariable(this.envVarApiKey);
     this.searchEngineId =
-      fields.searchEngineId ?? getEnvironmentVariable(this.envVarSearchEngineId);
+      fields[this.envVarSearchEngineId] ?? getEnvironmentVariable(this.envVarSearchEngineId);
+
+    if (!this.override && (!this.apiKey || !this.searchEngineId)) {
+      throw new Error(
+        `Missing ${this.envVarApiKey} or ${this.envVarSearchEngineId} environment variable.`,
+      );
+    }
 
     this.kwargs = fields?.kwargs ?? {};
     this.name = 'google';

api/app/clients/tools/structured/StableDiffusion.js

@@ -80,13 +80,18 @@ class StableDiffusionAPI extends StructuredTool {
     const payload = {
       prompt,
       negative_prompt,
-      sampler_index: 'DPM++ 2M Karras',
       cfg_scale: 4.5,
       steps: 22,
       width: 1024,
       height: 1024,
     };
-    const generationResponse = await axios.post(`${url}/sdapi/v1/txt2img`, payload);
+    let generationResponse;
+    try {
+      generationResponse = await axios.post(`${url}/sdapi/v1/txt2img`, payload);
+    } catch (error) {
+      logger.error('[StableDiffusion] Error while generating image:', error);
+      return 'Error making API request.';
+    }
     const image = generationResponse.data.images[0];
 
     /** @type {{ height: number, width: number, seed: number, infotexts: string[] }} */

api/app/clients/tools/structured/TavilySearchResults.js

@@ -12,7 +12,7 @@ class TavilySearchResults extends Tool {
     this.envVar = 'TAVILY_API_KEY';
     /* Used to initialize the Tool without necessary variables. */
     this.override = fields.override ?? false;
-    this.apiKey = fields.apiKey ?? this.getApiKey();
+    this.apiKey = fields[this.envVar] ?? this.getApiKey();
 
     this.kwargs = fields?.kwargs ?? {};
     this.name = 'tavily_search_results_json';
@@ -82,7 +82,9 @@ class TavilySearchResults extends Tool {
 
     const json = await response.json();
     if (!response.ok) {
-      throw new Error(`Request failed with status ${response.status}: ${json.error}`);
+      throw new Error(
+        `Request failed with status ${response.status}: ${json?.detail?.error || json?.error}`,
+      );
     }
 
     return JSON.stringify(json);

api/app/clients/tools/structured/specs/GoogleSearch.spec.js

@@ -0,0 +1,50 @@
+const GoogleSearch = require('../GoogleSearch');
+
+jest.mock('node-fetch');
+jest.mock('@langchain/core/utils/env');
+
+describe('GoogleSearch', () => {
+  let originalEnv;
+  const mockApiKey = 'mock_api';
+  const mockSearchEngineId = 'mock_search_engine_id';
+
+  beforeAll(() => {
+    originalEnv = { ...process.env };
+  });
+
+  beforeEach(() => {
+    jest.resetModules();
+    process.env = {
+      ...originalEnv,
+      GOOGLE_SEARCH_API_KEY: mockApiKey,
+      GOOGLE_CSE_ID: mockSearchEngineId,
+    };
+  });
+
+  afterEach(() => {
+    jest.clearAllMocks();
+    process.env = originalEnv;
+  });
+
+  it('should use mockApiKey and mockSearchEngineId when environment variables are not set', () => {
+    const instance = new GoogleSearch({
+      GOOGLE_SEARCH_API_KEY: mockApiKey,
+      GOOGLE_CSE_ID: mockSearchEngineId,
+    });
+    expect(instance.apiKey).toBe(mockApiKey);
+    expect(instance.searchEngineId).toBe(mockSearchEngineId);
+  });
+
+  it('should throw an error if GOOGLE_SEARCH_API_KEY or GOOGLE_CSE_ID is missing', () => {
+    delete process.env.GOOGLE_SEARCH_API_KEY;
+    expect(() => new GoogleSearch()).toThrow(
+      'Missing GOOGLE_SEARCH_API_KEY or GOOGLE_CSE_ID environment variable.',
+    );
+
+    process.env.GOOGLE_SEARCH_API_KEY = mockApiKey;
+    delete process.env.GOOGLE_CSE_ID;
+    expect(() => new GoogleSearch()).toThrow(
+      'Missing GOOGLE_SEARCH_API_KEY or GOOGLE_CSE_ID environment variable.',
+    );
+  });
+});

api/app/clients/tools/structured/specs/TavilySearchResults.spec.js

@@ -0,0 +1,38 @@
+const TavilySearchResults = require('../TavilySearchResults');
+
+jest.mock('node-fetch');
+jest.mock('@langchain/core/utils/env');
+
+describe('TavilySearchResults', () => {
+  let originalEnv;
+  const mockApiKey = 'mock_api_key';
+
+  beforeAll(() => {
+    originalEnv = { ...process.env };
+  });
+
+  beforeEach(() => {
+    jest.resetModules();
+    process.env = {
+      ...originalEnv,
+      TAVILY_API_KEY: mockApiKey,
+    };
+  });
+
+  afterEach(() => {
+    jest.clearAllMocks();
+    process.env = originalEnv;
+  });
+
+  it('should throw an error if TAVILY_API_KEY is missing', () => {
+    delete process.env.TAVILY_API_KEY;
+    expect(() => new TavilySearchResults()).toThrow('Missing TAVILY_API_KEY environment variable.');
+  });
+
+  it('should use mockApiKey when TAVILY_API_KEY is not set in the environment', () => {
+    const instance = new TavilySearchResults({
+      TAVILY_API_KEY: mockApiKey,
+    });
+    expect(instance.apiKey).toBe(mockApiKey);
+  });
+});

api/cache/clearPendingReq.js

@@ -35,7 +35,7 @@ const clearPendingReq = async ({ userId, cache: _cache }) => {
     return;
   }
 
-  const key = `${USE_REDIS ? namespace : ''}:${userId ?? ''}`;
+  const key = `${isEnabled(USE_REDIS) ? namespace : ''}:${userId ?? ''}`;
   const currentReq = +((await cache.get(key)) ?? 0);
 
   if (currentReq && currentReq >= 1) {

api/cache/getLogStores.js

@@ -1,12 +1,11 @@
 const Keyv = require('keyv');
-const { CacheKeys, ViolationTypes } = require('librechat-data-provider');
+const { CacheKeys, ViolationTypes, Time } = require('librechat-data-provider');
 const { logFile, violationFile } = require('./keyvFiles');
 const { math, isEnabled } = require('~/server/utils');
 const keyvRedis = require('./keyvRedis');
 const keyvMongo = require('./keyvMongo');
 
 const { BAN_DURATION, USE_REDIS } = process.env ?? {};
-const THIRTY_MINUTES = 1800000;
 
 const duration = math(BAN_DURATION, 7200000);
 
@@ -24,13 +23,25 @@ const config = isEnabled(USE_REDIS)
   ? new Keyv({ store: keyvRedis })
   : new Keyv({ namespace: CacheKeys.CONFIG_STORE });
 
-const tokenConfig = isEnabled(USE_REDIS) // ttl: 30 minutes
-  ? new Keyv({ store: keyvRedis, ttl: THIRTY_MINUTES })
-  : new Keyv({ namespace: CacheKeys.TOKEN_CONFIG, ttl: THIRTY_MINUTES });
+const roles = isEnabled(USE_REDIS)
+  ? new Keyv({ store: keyvRedis })
+  : new Keyv({ namespace: CacheKeys.ROLES });
 
-const genTitle = isEnabled(USE_REDIS) // ttl: 2 minutes
-  ? new Keyv({ store: keyvRedis, ttl: 120000 })
-  : new Keyv({ namespace: CacheKeys.GEN_TITLE, ttl: 120000 });
+const audioRuns = isEnabled(USE_REDIS)
+  ? new Keyv({ store: keyvRedis, ttl: Time.TEN_MINUTES })
+  : new Keyv({ namespace: CacheKeys.AUDIO_RUNS, ttl: Time.TEN_MINUTES });
+
+const messages = isEnabled(USE_REDIS)
+  ? new Keyv({ store: keyvRedis, ttl: Time.FIVE_MINUTES })
+  : new Keyv({ namespace: CacheKeys.MESSAGES, ttl: Time.FIVE_MINUTES });
+
+const tokenConfig = isEnabled(USE_REDIS)
+  ? new Keyv({ store: keyvRedis, ttl: Time.THIRTY_MINUTES })
+  : new Keyv({ namespace: CacheKeys.TOKEN_CONFIG, ttl: Time.THIRTY_MINUTES });
+
+const genTitle = isEnabled(USE_REDIS)
+  ? new Keyv({ store: keyvRedis, ttl: Time.TWO_MINUTES })
+  : new Keyv({ namespace: CacheKeys.GEN_TITLE, ttl: Time.TWO_MINUTES });
 
 const modelQueries = isEnabled(process.env.USE_REDIS)
   ? new Keyv({ store: keyvRedis })
@@ -38,9 +49,10 @@ const modelQueries = isEnabled(process.env.USE_REDIS)
 
 const abortKeys = isEnabled(USE_REDIS)
   ? new Keyv({ store: keyvRedis })
-  : new Keyv({ namespace: CacheKeys.ABORT_KEYS, ttl: 600000 });
+  : new Keyv({ namespace: CacheKeys.ABORT_KEYS, ttl: Time.TEN_MINUTES });
 
 const namespaces = {
+  [CacheKeys.ROLES]: roles,
   [CacheKeys.CONFIG_STORE]: config,
   pending_req,
   [ViolationTypes.BAN]: new Keyv({ store: keyvMongo, namespace: CacheKeys.BANS, ttl: duration }),
@@ -55,7 +67,14 @@ const namespaces = {
   message_limit: createViolationInstance('message_limit'),
   token_balance: createViolationInstance(ViolationTypes.TOKEN_BALANCE),
   registrations: createViolationInstance('registrations'),
+  [ViolationTypes.TTS_LIMIT]: createViolationInstance(ViolationTypes.TTS_LIMIT),
+  [ViolationTypes.STT_LIMIT]: createViolationInstance(ViolationTypes.STT_LIMIT),
+  [ViolationTypes.CONVO_ACCESS]: createViolationInstance(ViolationTypes.CONVO_ACCESS),
   [ViolationTypes.FILE_UPLOAD_LIMIT]: createViolationInstance(ViolationTypes.FILE_UPLOAD_LIMIT),
+  [ViolationTypes.VERIFY_EMAIL_LIMIT]: createViolationInstance(ViolationTypes.VERIFY_EMAIL_LIMIT),
+  [ViolationTypes.RESET_PASSWORD_LIMIT]: createViolationInstance(
+    ViolationTypes.RESET_PASSWORD_LIMIT,
+  ),
   [ViolationTypes.ILLEGAL_MODEL_REQUEST]: createViolationInstance(
     ViolationTypes.ILLEGAL_MODEL_REQUEST,
   ),
@@ -64,6 +83,8 @@ const namespaces = {
   [CacheKeys.TOKEN_CONFIG]: tokenConfig,
   [CacheKeys.GEN_TITLE]: genTitle,
   [CacheKeys.MODEL_QUERIES]: modelQueries,
+  [CacheKeys.AUDIO_RUNS]: audioRuns,
+  [CacheKeys.MESSAGES]: messages,
 };
 
 /**

api/cache/logViolation.js

@@ -1,6 +1,6 @@
+const { isEnabled } = require('~/server/utils');
 const getLogStores = require('./getLogStores');
 const banViolation = require('./banViolation');
-const { isEnabled } = require('../server/utils');
 
 /**
  * Logs the violation.

api/config/parsers.js

@@ -27,26 +27,25 @@ function getMatchingSensitivePatterns(valueStr) {
 }
 
 /**
- * Redacts sensitive information from a console message.
- *
+ * Redacts sensitive information from a console message and trims it to a specified length if provided.
  * @param {string} str - The console message to be redacted.
- * @returns {string} - The redacted console message.
+ * @param {number} [trimLength] - The optional length at which to trim the redacted message.
+ * @returns {string} - The redacted and optionally trimmed console message.
  */
-function redactMessage(str) {
+function redactMessage(str, trimLength) {
   if (!str) {
     return '';
   }
 
   const patterns = getMatchingSensitivePatterns(str);
-
-  if (patterns.length === 0) {
-    return str;
-  }
-
   patterns.forEach((pattern) => {
     str = str.replace(pattern, '$1[REDACTED]');
   });
 
+  if (trimLength !== undefined && str.length > trimLength) {
+    return `${str.substring(0, trimLength)}...`;
+  }
+
   return str;
 }
 
@@ -110,6 +109,14 @@ const condenseArray = (item) => {
  * @returns {string} - The formatted log message.
  */
 const debugTraverse = winston.format.printf(({ level, message, timestamp, ...metadata }) => {
+  if (!message) {
+    return `${timestamp} ${level}`;
+  }
+
+  if (!message?.trim || typeof message !== 'string') {
+    return `${timestamp} ${level}: ${JSON.stringify(message)}`;
+  }
+
   let msg = `${timestamp} ${level}: ${truncateLongStrings(message?.trim(), 150)}`;
   try {
     if (level !== 'debug') {

api/models/Action.js

@@ -62,8 +62,24 @@ const deleteAction = async (searchParams, session = null) => {
   return await Action.findOneAndDelete(searchParams, options).lean();
 };
 
-module.exports = {
-  updateAction,
-  getActions,
-  deleteAction,
+/**
+ * Deletes actions by params, within a transaction session if provided.
+ *
+ * @param {Object} searchParams - The search parameters to find the actions to delete.
+ * @param {string} searchParams.action_id - The ID of the action(s) to delete.
+ * @param {string} searchParams.user - The user ID of the action's author.
+ * @param {mongoose.ClientSession} [session] - The transaction session to use (optional).
+ * @returns {Promise<Number>} A promise that resolves to the number of deleted action documents.
+ */
+const deleteActions = async (searchParams, session = null) => {
+  const options = session ? { session } : {};
+  const result = await Action.deleteMany(searchParams, options);
+  return result.deletedCount;
+};
+
+module.exports = {
+  getActions,
+  updateAction,
+  deleteAction,
+  deleteActions,
 };

api/models/Assistant.js

@@ -14,7 +14,7 @@ const Assistant = mongoose.model('assistant', assistantSchema);
  * @param {mongoose.ClientSession} [session] - The transaction session to use (optional).
  * @returns {Promise<Object>} The updated or newly created assistant document as a plain object.
  */
-const updateAssistant = async (searchParams, updateData, session = null) => {
+const updateAssistantDoc = async (searchParams, updateData, session = null) => {
   const options = { new: true, upsert: true, session };
   return await Assistant.findOneAndUpdate(searchParams, updateData, options).lean();
 };
@@ -39,8 +39,21 @@ const getAssistants = async (searchParams) => {
   return await Assistant.find(searchParams).lean();
 };
 
+/**
+ * Deletes an assistant based on the provided ID.
+ *
+ * @param {Object} searchParams - The search parameters to find the assistant to delete.
+ * @param {string} searchParams.assistant_id - The ID of the assistant to delete.
+ * @param {string} searchParams.user - The user ID of the assistant's author.
+ * @returns {Promise<void>} Resolves when the assistant has been successfully deleted.
+ */
+const deleteAssistant = async (searchParams) => {
+  return await Assistant.findOneAndDelete(searchParams);
+};
+
 module.exports = {
-  updateAssistant,
+  updateAssistantDoc,
+  deleteAssistant,
   getAssistants,
   getAssistant,
 };

api/models/Categories.js

@@ -0,0 +1,61 @@
+const { logger } = require('~/config');
+// const { Categories } = require('./schema/categories');
+const options = [
+  {
+    label: '',
+    value: '',
+  },
+  {
+    label: 'idea',
+    value: 'idea',
+  },
+  {
+    label: 'travel',
+    value: 'travel',
+  },
+  {
+    label: 'teach_or_explain',
+    value: 'teach_or_explain',
+  },
+  {
+    label: 'write',
+    value: 'write',
+  },
+  {
+    label: 'shop',
+    value: 'shop',
+  },
+  {
+    label: 'code',
+    value: 'code',
+  },
+  {
+    label: 'misc',
+    value: 'misc',
+  },
+  {
+    label: 'roleplay',
+    value: 'roleplay',
+  },
+  {
+    label: 'finance',
+    value: 'finance',
+  },
+];
+
+module.exports = {
+  /**
+   * Retrieves the categories asynchronously.
+   * @returns {Promise<TGetCategoriesResponse>} An array of category objects.
+   * @throws {Error} If there is an error retrieving the categories.
+   */
+  getCategories: async () => {
+    try {
+      // const categories = await Categories.find();
+      return options;
+    } catch (error) {
+      logger.error('Error getting categories', error);
+      return [];
+    }
+  },
+};

api/models/Conversation.js

@@ -2,6 +2,20 @@ const Conversation = require('./schema/convoSchema');
 const { getMessages, deleteMessages } = require('./Message');
 const logger = require('~/config/winston');
 
+/**
+ * Searches for a conversation by conversationId and returns a lean document with only conversationId and user.
+ * @param {string} conversationId - The conversation's ID.
+ * @returns {Promise<{conversationId: string, user: string} | null>} The conversation object with selected fields or null if not found.
+ */
+const searchConversation = async (conversationId) => {
+  try {
+    return await Conversation.findOne({ conversationId }, 'conversationId user').lean();
+  } catch (error) {
+    logger.error('[searchConversation] Error searching conversation', error);
+    throw new Error('Error searching conversation');
+  }
+};
+
 /**
  * Retrieves a single conversation for a given user and conversation ID.
  * @param {string} user - The user's ID.
@@ -19,20 +33,40 @@ const getConvo = async (user, conversationId) => {
 
 module.exports = {
   Conversation,
-  saveConvo: async (user, { conversationId, newConversationId, ...convo }) => {
+  searchConversation,
+  /**
+   * Saves a conversation to the database.
+   * @param {Object} req - The request object.
+   * @param {string} conversationId - The conversation's ID.
+   * @param {Object} metadata - Additional metadata to log for operation.
+   * @returns {Promise<TConversation>} The conversation object.
+   */
+  saveConvo: async (req, { conversationId, newConversationId, ...convo }, metadata) => {
     try {
-      const messages = await getMessages({ conversationId });
-      const update = { ...convo, messages, user };
+      if (metadata && metadata?.context) {
+        logger.debug(`[saveConvo] ${metadata.context}`);
+      }
+      const messages = await getMessages({ conversationId }, '_id');
+      const update = { ...convo, messages, user: req.user.id };
       if (newConversationId) {
         update.conversationId = newConversationId;
       }
 
-      return await Conversation.findOneAndUpdate({ conversationId: conversationId, user }, update, {
-        new: true,
-        upsert: true,
-      });
+      const conversation = await Conversation.findOneAndUpdate(
+        { conversationId, user: req.user.id },
+        update,
+        {
+          new: true,
+          upsert: true,
+        },
+      );
+
+      return conversation.toObject();
     } catch (error) {
       logger.error('[saveConvo] Error saving conversation', error);
+      if (metadata && metadata?.context) {
+        logger.info(`[saveConvo] ${metadata.context}`);
+      }
       return { message: 'Error saving conversation' };
     }
   },
@@ -54,13 +88,16 @@ module.exports = {
       throw new Error('Failed to save conversations in bulk.');
     }
   },
-  getConvosByPage: async (user, pageNumber = 1, pageSize = 25, isArchived = false) => {
+  getConvosByPage: async (user, pageNumber = 1, pageSize = 25, isArchived = false, tags) => {
     const query = { user };
     if (isArchived) {
       query.isArchived = true;
     } else {
       query.$or = [{ isArchived: false }, { isArchived: { $exists: false } }];
     }
+    if (Array.isArray(tags) && tags.length > 0) {
+      query.tags = { $in: tags };
+    }
     try {
       const totalConvos = (await Conversation.countDocuments(query)) || 1;
       const totalPages = Math.ceil(totalConvos / pageSize);

api/models/ConversationTag.js

@@ -0,0 +1,249 @@
+const ConversationTag = require('./schema/conversationTagSchema');
+const Conversation = require('./schema/convoSchema');
+const logger = require('~/config/winston');
+
+/**
+ * Retrieves all conversation tags for a user.
+ * @param {string} user - The user ID.
+ * @returns {Promise<Array>} An array of conversation tags.
+ */
+const getConversationTags = async (user) => {
+  try {
+    return await ConversationTag.find({ user }).sort({ position: 1 }).lean();
+  } catch (error) {
+    logger.error('[getConversationTags] Error getting conversation tags', error);
+    throw new Error('Error getting conversation tags');
+  }
+};
+
+/**
+ * Creates a new conversation tag.
+ * @param {string} user - The user ID.
+ * @param {Object} data - The tag data.
+ * @param {string} data.tag - The tag name.
+ * @param {string} [data.description] - The tag description.
+ * @param {boolean} [data.addToConversation] - Whether to add the tag to a conversation.
+ * @param {string} [data.conversationId] - The conversation ID to add the tag to.
+ * @returns {Promise<Object>} The created tag.
+ */
+const createConversationTag = async (user, data) => {
+  try {
+    const { tag, description, addToConversation, conversationId } = data;
+
+    const existingTag = await ConversationTag.findOne({ user, tag }).lean();
+    if (existingTag) {
+      return existingTag;
+    }
+
+    const maxPosition = await ConversationTag.findOne({ user }).sort('-position').lean();
+    const position = (maxPosition?.position || 0) + 1;
+
+    const newTag = await ConversationTag.findOneAndUpdate(
+      { tag, user },
+      {
+        tag,
+        user,
+        count: addToConversation ? 1 : 0,
+        position,
+        description,
+        $setOnInsert: { createdAt: new Date() },
+      },
+      {
+        new: true,
+        upsert: true,
+        lean: true,
+      },
+    );
+
+    if (addToConversation && conversationId) {
+      await Conversation.findOneAndUpdate(
+        { user, conversationId },
+        { $addToSet: { tags: tag } },
+        { new: true },
+      );
+    }
+
+    return newTag;
+  } catch (error) {
+    logger.error('[createConversationTag] Error creating conversation tag', error);
+    throw new Error('Error creating conversation tag');
+  }
+};
+
+/**
+ * Updates an existing conversation tag.
+ * @param {string} user - The user ID.
+ * @param {string} oldTag - The current tag name.
+ * @param {Object} data - The updated tag data.
+ * @param {string} [data.tag] - The new tag name.
+ * @param {string} [data.description] - The updated description.
+ * @param {number} [data.position] - The new position.
+ * @returns {Promise<Object>} The updated tag.
+ */
+const updateConversationTag = async (user, oldTag, data) => {
+  try {
+    const { tag: newTag, description, position } = data;
+
+    const existingTag = await ConversationTag.findOne({ user, tag: oldTag }).lean();
+    if (!existingTag) {
+      return null;
+    }
+
+    if (newTag && newTag !== oldTag) {
+      const tagAlreadyExists = await ConversationTag.findOne({ user, tag: newTag }).lean();
+      if (tagAlreadyExists) {
+        throw new Error('Tag already exists');
+      }
+
+      await Conversation.updateMany({ user, tags: oldTag }, { $set: { 'tags.$': newTag } });
+    }
+
+    const updateData = {};
+    if (newTag) {
+      updateData.tag = newTag;
+    }
+    if (description !== undefined) {
+      updateData.description = description;
+    }
+    if (position !== undefined) {
+      await adjustPositions(user, existingTag.position, position);
+      updateData.position = position;
+    }
+
+    return await ConversationTag.findOneAndUpdate({ user, tag: oldTag }, updateData, {
+      new: true,
+      lean: true,
+    });
+  } catch (error) {
+    logger.error('[updateConversationTag] Error updating conversation tag', error);
+    throw new Error('Error updating conversation tag');
+  }
+};
+
+/**
+ * Adjusts positions of tags when a tag's position is changed.
+ * @param {string} user - The user ID.
+ * @param {number} oldPosition - The old position of the tag.
+ * @param {number} newPosition - The new position of the tag.
+ * @returns {Promise<void>}
+ */
+const adjustPositions = async (user, oldPosition, newPosition) => {
+  if (oldPosition === newPosition) {
+    return;
+  }
+
+  const update = oldPosition < newPosition ? { $inc: { position: -1 } } : { $inc: { position: 1 } };
+  const position =
+    oldPosition < newPosition
+      ? {
+        $gt: Math.min(oldPosition, newPosition),
+        $lte: Math.max(oldPosition, newPosition),
+      }
+      : {
+        $gte: Math.min(oldPosition, newPosition),
+        $lt: Math.max(oldPosition, newPosition),
+      };
+
+  await ConversationTag.updateMany(
+    {
+      user,
+      position,
+    },
+    update,
+  );
+};
+
+/**
+ * Deletes a conversation tag.
+ * @param {string} user - The user ID.
+ * @param {string} tag - The tag to delete.
+ * @returns {Promise<Object>} The deleted tag.
+ */
+const deleteConversationTag = async (user, tag) => {
+  try {
+    const deletedTag = await ConversationTag.findOneAndDelete({ user, tag }).lean();
+    if (!deletedTag) {
+      return null;
+    }
+
+    await Conversation.updateMany({ user, tags: tag }, { $pull: { tags: tag } });
+
+    await ConversationTag.updateMany(
+      { user, position: { $gt: deletedTag.position } },
+      { $inc: { position: -1 } },
+    );
+
+    return deletedTag;
+  } catch (error) {
+    logger.error('[deleteConversationTag] Error deleting conversation tag', error);
+    throw new Error('Error deleting conversation tag');
+  }
+};
+
+/**
+ * Updates tags for a specific conversation.
+ * @param {string} user - The user ID.
+ * @param {string} conversationId - The conversation ID.
+ * @param {string[]} tags - The new set of tags for the conversation.
+ * @returns {Promise<string[]>} The updated list of tags for the conversation.
+ */
+const updateTagsForConversation = async (user, conversationId, tags) => {
+  try {
+    const conversation = await Conversation.findOne({ user, conversationId }).lean();
+    if (!conversation) {
+      throw new Error('Conversation not found');
+    }
+
+    const oldTags = new Set(conversation.tags);
+    const newTags = new Set(tags);
+
+    const addedTags = [...newTags].filter((tag) => !oldTags.has(tag));
+    const removedTags = [...oldTags].filter((tag) => !newTags.has(tag));
+
+    const bulkOps = [];
+
+    for (const tag of addedTags) {
+      bulkOps.push({
+        updateOne: {
+          filter: { user, tag },
+          update: { $inc: { count: 1 } },
+          upsert: true,
+        },
+      });
+    }
+
+    for (const tag of removedTags) {
+      bulkOps.push({
+        updateOne: {
+          filter: { user, tag },
+          update: { $inc: { count: -1 } },
+        },
+      });
+    }
+
+    if (bulkOps.length > 0) {
+      await ConversationTag.bulkWrite(bulkOps);
+    }
+
+    const updatedConversation = (
+      await Conversation.findOneAndUpdate(
+        { user, conversationId },
+        { $set: { tags: [...newTags] } },
+        { new: true },
+      )
+    ).toObject();
+
+    return updatedConversation.tags;
+  } catch (error) {
+    logger.error('[updateTagsForConversation] Error updating tags', error);
+    throw new Error('Error updating tags for conversation');
+  }
+};
+
+module.exports = {
+  getConversationTags,
+  createConversationTag,
+  updateConversationTag,
+  deleteConversationTag,
+  updateTagsForConversation,
+};

api/models/File.js

@@ -97,8 +97,12 @@ const deleteFileByFilter = async (filter) => {
  * @param {Array<string>} file_ids - The unique identifiers of the files to delete.
  * @returns {Promise<Object>} A promise that resolves to the result of the deletion operation.
  */
-const deleteFiles = async (file_ids) => {
-  return await File.deleteMany({ file_id: { $in: file_ids } });
+const deleteFiles = async (file_ids, user) => {
+  let deleteQuery = { file_id: { $in: file_ids } };
+  if (user) {
+    deleteQuery = { user: user };
+  }
+  return await File.deleteMany(deleteQuery);
 };
 
 module.exports = {

api/models/Message.js

@@ -1,191 +1,342 @@
 const { z } = require('zod');
 const Message = require('./schema/messageSchema');
-const logger = require('~/config/winston');
+const { logger } = require('~/config');
 
 const idSchema = z.string().uuid();
 
+/**
+ * Saves a message in the database.
+ *
+ * @async
+ * @function saveMessage
+ * @param {Express.Request} req - The request object containing user information.
+ * @param {Object} params - The message data object.
+ * @param {string} params.endpoint - The endpoint where the message originated.
+ * @param {string} params.iconURL - The URL of the sender's icon.
+ * @param {string} params.messageId - The unique identifier for the message.
+ * @param {string} params.newMessageId - The new unique identifier for the message (if applicable).
+ * @param {string} params.conversationId - The identifier of the conversation.
+ * @param {string} [params.parentMessageId] - The identifier of the parent message, if any.
+ * @param {string} params.sender - The identifier of the sender.
+ * @param {string} params.text - The text content of the message.
+ * @param {boolean} params.isCreatedByUser - Indicates if the message was created by the user.
+ * @param {string} [params.error] - Any error associated with the message.
+ * @param {boolean} [params.unfinished] - Indicates if the message is unfinished.
+ * @param {Object[]} [params.files] - An array of files associated with the message.
+ * @param {boolean} [params.isEdited] - Indicates if the message was edited.
+ * @param {string} [params.finish_reason] - Reason for finishing the message.
+ * @param {number} [params.tokenCount] - The number of tokens in the message.
+ * @param {string} [params.plugin] - Plugin associated with the message.
+ * @param {string[]} [params.plugins] - An array of plugins associated with the message.
+ * @param {string} [params.model] - The model used to generate the message.
+ * @param {Object} [metadata] - Additional metadata for this operation
+ * @param {string} [metadata.context] - The context of the operation
+ * @returns {Promise<TMessage>} The updated or newly inserted message document.
+ * @throws {Error} If there is an error in saving the message.
+ */
+async function saveMessage(req, params, metadata) {
+  try {
+    if (!req || !req.user || !req.user.id) {
+      throw new Error('User not authenticated');
+    }
+
+    const {
+      text,
+      error,
+      model,
+      files,
+      plugin,
+      sender,
+      plugins,
+      iconURL,
+      endpoint,
+      isEdited,
+      messageId,
+      unfinished,
+      tokenCount,
+      newMessageId,
+      finish_reason,
+      conversationId,
+      parentMessageId,
+      isCreatedByUser,
+    } = params;
+
+    const validConvoId = idSchema.safeParse(conversationId);
+    if (!validConvoId.success) {
+      logger.warn(`Invalid conversation ID: ${conversationId}`);
+      if (metadata && metadata?.context) {
+        logger.info(`---\`saveMessage\` context: ${metadata.context}`);
+      }
+
+      logger.info(`---Invalid conversation ID Params:
+
+${JSON.stringify(params, null, 2)}
+
+`);
+      return;
+    }
+
+    const update = {
+      user: req.user.id,
+      iconURL,
+      endpoint,
+      messageId: newMessageId || messageId,
+      conversationId,
+      parentMessageId,
+      sender,
+      text,
+      isCreatedByUser,
+      isEdited,
+      finish_reason,
+      error,
+      unfinished,
+      tokenCount,
+      plugin,
+      plugins,
+      model,
+    };
+
+    if (files) {
+      update.files = files;
+    }
+
+    const message = await Message.findOneAndUpdate({ messageId, user: req.user.id }, update, {
+      upsert: true,
+      new: true,
+    });
+
+    return message.toObject();
+  } catch (err) {
+    logger.error('Error saving message:', err);
+    if (metadata && metadata?.context) {
+      logger.info(`---\`saveMessage\` context: ${metadata.context}`);
+    }
+    throw err;
+  }
+}
+
+/**
+ * Saves multiple messages in the database in bulk.
+ *
+ * @async
+ * @function bulkSaveMessages
+ * @param {Object[]} messages - An array of message objects to save.
+ * @returns {Promise<Object>} The result of the bulk write operation.
+ * @throws {Error} If there is an error in saving messages in bulk.
+ */
+async function bulkSaveMessages(messages) {
+  try {
+    const bulkOps = messages.map((message) => ({
+      updateOne: {
+        filter: { messageId: message.messageId },
+        update: message,
+        upsert: true,
+      },
+    }));
+
+    const result = await Message.bulkWrite(bulkOps);
+    return result;
+  } catch (err) {
+    logger.error('Error saving messages in bulk:', err);
+    throw err;
+  }
+}
+
+/**
+ * Records a message in the database.
+ *
+ * @async
+ * @function recordMessage
+ * @param {Object} params - The message data object.
+ * @param {string} params.user - The identifier of the user.
+ * @param {string} params.endpoint - The endpoint where the message originated.
+ * @param {string} params.messageId - The unique identifier for the message.
+ * @param {string} params.conversationId - The identifier of the conversation.
+ * @param {string} [params.parentMessageId] - The identifier of the parent message, if any.
+ * @param {Partial<TMessage>} rest - Any additional properties from the TMessage typedef not explicitly listed.
+ * @returns {Promise<Object>} The updated or newly inserted message document.
+ * @throws {Error} If there is an error in saving the message.
+ */
+async function recordMessage({
+  user,
+  endpoint,
+  messageId,
+  conversationId,
+  parentMessageId,
+  ...rest
+}) {
+  try {
+    // No parsing of convoId as may use threadId
+    const message = {
+      user,
+      endpoint,
+      messageId,
+      conversationId,
+      parentMessageId,
+      ...rest,
+    };
+
+    return await Message.findOneAndUpdate({ user, messageId }, message, {
+      upsert: true,
+      new: true,
+    });
+  } catch (err) {
+    logger.error('Error recording message:', err);
+    throw err;
+  }
+}
+
+/**
+ * Updates the text of a message.
+ *
+ * @async
+ * @function updateMessageText
+ * @param {Object} params - The update data object.
+ * @param {Object} req - The request object.
+ * @param {string} params.messageId - The unique identifier for the message.
+ * @param {string} params.text - The new text content of the message.
+ * @returns {Promise<void>}
+ * @throws {Error} If there is an error in updating the message text.
+ */
+async function updateMessageText(req, { messageId, text }) {
+  try {
+    await Message.updateOne({ messageId, user: req.user.id }, { text });
+  } catch (err) {
+    logger.error('Error updating message text:', err);
+    throw err;
+  }
+}
+
+/**
+ * Updates a message.
+ *
+ * @async
+ * @function updateMessage
+ * @param {Object} req - The request object.
+ * @param {Object} message - The message object containing update data.
+ * @param {string} message.messageId - The unique identifier for the message.
+ * @param {string} [message.text] - The new text content of the message.
+ * @param {Object[]} [message.files] - The files associated with the message.
+ * @param {boolean} [message.isCreatedByUser] - Indicates if the message was created by the user.
+ * @param {string} [message.sender] - The identifier of the sender.
+ * @param {number} [message.tokenCount] - The number of tokens in the message.
+ * @param {Object} [metadata] - The operation metadata
+ * @param {string} [metadata.context] - The operation metadata
+ * @returns {Promise<TMessage>} The updated message document.
+ * @throws {Error} If there is an error in updating the message or if the message is not found.
+ */
+async function updateMessage(req, message, metadata) {
+  try {
+    const { messageId, ...update } = message;
+    update.isEdited = true;
+    const updatedMessage = await Message.findOneAndUpdate(
+      { messageId, user: req.user.id },
+      update,
+      {
+        new: true,
+      },
+    );
+
+    if (!updatedMessage) {
+      throw new Error('Message not found or user not authorized.');
+    }
+
+    return {
+      messageId: updatedMessage.messageId,
+      conversationId: updatedMessage.conversationId,
+      parentMessageId: updatedMessage.parentMessageId,
+      sender: updatedMessage.sender,
+      text: updatedMessage.text,
+      isCreatedByUser: updatedMessage.isCreatedByUser,
+      tokenCount: updatedMessage.tokenCount,
+      isEdited: true,
+    };
+  } catch (err) {
+    logger.error('Error updating message:', err);
+    if (metadata && metadata?.context) {
+      logger.info(`---\`updateMessage\` context: ${metadata.context}`);
+    }
+    throw err;
+  }
+}
+
+/**
+ * Deletes messages in a conversation since a specific message.
+ *
+ * @async
+ * @function deleteMessagesSince
+ * @param {Object} params - The parameters object.
+ * @param {Object} req - The request object.
+ * @param {string} params.messageId - The unique identifier for the message.
+ * @param {string} params.conversationId - The identifier of the conversation.
+ * @returns {Promise<Number>} The number of deleted messages.
+ * @throws {Error} If there is an error in deleting messages.
+ */
+async function deleteMessagesSince(req, { messageId, conversationId }) {
+  try {
+    const message = await Message.findOne({ messageId, user: req.user.id }).lean();
+
+    if (message) {
+      const query = Message.find({ conversationId, user: req.user.id });
+      return await query.deleteMany({
+        createdAt: { $gt: message.createdAt },
+      });
+    }
+    return undefined;
+  } catch (err) {
+    logger.error('Error deleting messages:', err);
+    throw err;
+  }
+}
+
+/**
+ * Retrieves messages from the database.
+ * @async
+ * @function getMessages
+ * @param {Record<string, unknown>} filter - The filter criteria.
+ * @param {string | undefined} [select] - The fields to select.
+ * @returns {Promise<TMessage[]>} The messages that match the filter criteria.
+ * @throws {Error} If there is an error in retrieving messages.
+ */
+async function getMessages(filter, select) {
+  try {
+    if (select) {
+      return await Message.find(filter).select(select).sort({ createdAt: 1 }).lean();
+    }
+
+    return await Message.find(filter).sort({ createdAt: 1 }).lean();
+  } catch (err) {
+    logger.error('Error getting messages:', err);
+    throw err;
+  }
+}
+
+/**
+ * Deletes messages from the database.
+ *
+ * @async
+ * @function deleteMessages
+ * @param {Object} filter - The filter criteria to find messages to delete.
+ * @returns {Promise<Object>} The metadata with count of deleted messages.
+ * @throws {Error} If there is an error in deleting messages.
+ */
+async function deleteMessages(filter) {
+  try {
+    return await Message.deleteMany(filter);
+  } catch (err) {
+    logger.error('Error deleting messages:', err);
+    throw err;
+  }
+}
+
 module.exports = {
   Message,
-
-  async saveMessage({
-    user,
-    endpoint,
-    iconURL,
-    messageId,
-    newMessageId,
-    conversationId,
-    parentMessageId,
-    sender,
-    text,
-    isCreatedByUser,
-    error,
-    unfinished,
-    files,
-    isEdited,
-    finish_reason,
-    tokenCount,
-    plugin,
-    plugins,
-    model,
-  }) {
-    try {
-      const validConvoId = idSchema.safeParse(conversationId);
-      if (!validConvoId.success) {
-        return;
-      }
-
-      const update = {
-        user,
-        iconURL,
-        endpoint,
-        messageId: newMessageId || messageId,
-        conversationId,
-        parentMessageId,
-        sender,
-        text,
-        isCreatedByUser,
-        isEdited,
-        finish_reason,
-        error,
-        unfinished,
-        tokenCount,
-        plugin,
-        plugins,
-        model,
-      };
-
-      if (files) {
-        update.files = files;
-      }
-      // may also need to update the conversation here
-      await Message.findOneAndUpdate({ messageId }, update, { upsert: true, new: true });
-
-      return {
-        messageId,
-        conversationId,
-        parentMessageId,
-        sender,
-        text,
-        isCreatedByUser,
-        tokenCount,
-      };
-    } catch (err) {
-      logger.error('Error saving message:', err);
-      throw new Error('Failed to save message.');
-    }
-  },
-
-  async bulkSaveMessages(messages) {
-    try {
-      const bulkOps = messages.map((message) => ({
-        updateOne: {
-          filter: { messageId: message.messageId },
-          update: message,
-          upsert: true,
-        },
-      }));
-
-      const result = await Message.bulkWrite(bulkOps);
-      return result;
-    } catch (err) {
-      logger.error('Error saving messages in bulk:', err);
-      throw new Error('Failed to save messages in bulk.');
-    }
-  },
-
-  /**
-   * Records a message in the database.
-   *
-   * @async
-   * @function recordMessage
-   * @param {Object} params - The message data object.
-   * @param {string} params.user - The identifier of the user.
-   * @param {string} params.endpoint - The endpoint where the message originated.
-   * @param {string} params.messageId - The unique identifier for the message.
-   * @param {string} params.conversationId - The identifier of the conversation.
-   * @param {string} [params.parentMessageId] - The identifier of the parent message, if any.
-   * @param {Partial<TMessage>} rest - Any additional properties from the TMessage typedef not explicitly listed.
-   * @returns {Promise<Object>} The updated or newly inserted message document.
-   * @throws {Error} If there is an error in saving the message.
-   */
-  async recordMessage({ user, endpoint, messageId, conversationId, parentMessageId, ...rest }) {
-    try {
-      // No parsing of convoId as may use threadId
-      const message = {
-        user,
-        endpoint,
-        messageId,
-        conversationId,
-        parentMessageId,
-        ...rest,
-      };
-
-      return await Message.findOneAndUpdate({ user, messageId }, message, {
-        upsert: true,
-        new: true,
-      });
-    } catch (err) {
-      logger.error('Error saving message:', err);
-      throw new Error('Failed to save message.');
-    }
-  },
-  async updateMessage(message) {
-    try {
-      const { messageId, ...update } = message;
-      update.isEdited = true;
-      const updatedMessage = await Message.findOneAndUpdate({ messageId }, update, {
-        new: true,
-      });
-
-      if (!updatedMessage) {
-        throw new Error('Message not found.');
-      }
-
-      return {
-        messageId: updatedMessage.messageId,
-        conversationId: updatedMessage.conversationId,
-        parentMessageId: updatedMessage.parentMessageId,
-        sender: updatedMessage.sender,
-        text: updatedMessage.text,
-        isCreatedByUser: updatedMessage.isCreatedByUser,
-        tokenCount: updatedMessage.tokenCount,
-        isEdited: true,
-      };
-    } catch (err) {
-      logger.error('Error updating message:', err);
-      throw new Error('Failed to update message.');
-    }
-  },
-  async deleteMessagesSince({ messageId, conversationId }) {
-    try {
-      const message = await Message.findOne({ messageId }).lean();
-
-      if (message) {
-        return await Message.find({ conversationId }).deleteMany({
-          createdAt: { $gt: message.createdAt },
-        });
-      }
-    } catch (err) {
-      logger.error('Error deleting messages:', err);
-      throw new Error('Failed to delete messages.');
-    }
-  },
-
-  async getMessages(filter) {
-    try {
-      return await Message.find(filter).sort({ createdAt: 1 }).lean();
-    } catch (err) {
-      logger.error('Error getting messages:', err);
-      throw new Error('Failed to get messages.');
-    }
-  },
-
-  async deleteMessages(filter) {
-    try {
-      return await Message.deleteMany(filter);
-    } catch (err) {
-      logger.error('Error deleting messages:', err);
-      throw new Error('Failed to delete messages.');
-    }
-  },
+  saveMessage,
+  bulkSaveMessages,
+  recordMessage,
+  updateMessageText,
+  updateMessage,
+  deleteMessagesSince,
+  getMessages,
+  deleteMessages,
 };

api/models/Message.spec.js

@@ -0,0 +1,239 @@
+const mongoose = require('mongoose');
+const { v4: uuidv4 } = require('uuid');
+
+jest.mock('mongoose');
+
+const mockFindQuery = {
+  select: jest.fn().mockReturnThis(),
+  sort: jest.fn().mockReturnThis(),
+  lean: jest.fn().mockReturnThis(),
+  deleteMany: jest.fn().mockResolvedValue({ deletedCount: 1 }),
+};
+
+const mockSchema = {
+  findOneAndUpdate: jest.fn(),
+  updateOne: jest.fn(),
+  findOne: jest.fn(() => ({
+    lean: jest.fn(),
+  })),
+  find: jest.fn(() => mockFindQuery),
+  deleteMany: jest.fn(),
+};
+
+mongoose.model.mockReturnValue(mockSchema);
+
+jest.mock('~/models/schema/messageSchema', () => mockSchema);
+
+jest.mock('~/config/winston', () => ({
+  error: jest.fn(),
+}));
+
+const {
+  saveMessage,
+  getMessages,
+  updateMessage,
+  deleteMessages,
+  updateMessageText,
+  deleteMessagesSince,
+} = require('~/models/Message');
+
+describe('Message Operations', () => {
+  let mockReq;
+  let mockMessage;
+
+  beforeEach(() => {
+    jest.clearAllMocks();
+
+    mockReq = {
+      user: { id: 'user123' },
+    };
+
+    mockMessage = {
+      messageId: 'msg123',
+      conversationId: uuidv4(),
+      text: 'Hello, world!',
+      user: 'user123',
+    };
+
+    mockSchema.findOneAndUpdate.mockResolvedValue({
+      toObject: () => mockMessage,
+    });
+  });
+
+  describe('saveMessage', () => {
+    it('should save a message for an authenticated user', async () => {
+      const result = await saveMessage(mockReq, mockMessage);
+      expect(result).toEqual(mockMessage);
+      expect(mockSchema.findOneAndUpdate).toHaveBeenCalledWith(
+        { messageId: 'msg123', user: 'user123' },
+        expect.objectContaining({ user: 'user123' }),
+        expect.any(Object),
+      );
+    });
+
+    it('should throw an error for unauthenticated user', async () => {
+      mockReq.user = null;
+      await expect(saveMessage(mockReq, mockMessage)).rejects.toThrow('User not authenticated');
+    });
+
+    it('should throw an error for invalid conversation ID', async () => {
+      mockMessage.conversationId = 'invalid-id';
+      await expect(saveMessage(mockReq, mockMessage)).resolves.toBeUndefined();
+    });
+  });
+
+  describe('updateMessageText', () => {
+    it('should update message text for the authenticated user', async () => {
+      await updateMessageText(mockReq, { messageId: 'msg123', text: 'Updated text' });
+      expect(mockSchema.updateOne).toHaveBeenCalledWith(
+        { messageId: 'msg123', user: 'user123' },
+        { text: 'Updated text' },
+      );
+    });
+  });
+
+  describe('updateMessage', () => {
+    it('should update a message for the authenticated user', async () => {
+      mockSchema.findOneAndUpdate.mockResolvedValue(mockMessage);
+      const result = await updateMessage(mockReq, { messageId: 'msg123', text: 'Updated text' });
+      expect(result).toEqual(
+        expect.objectContaining({
+          messageId: 'msg123',
+          text: 'Hello, world!',
+          isEdited: true,
+        }),
+      );
+    });
+
+    it('should throw an error if message is not found', async () => {
+      mockSchema.findOneAndUpdate.mockResolvedValue(null);
+      await expect(
+        updateMessage(mockReq, { messageId: 'nonexistent', text: 'Test' }),
+      ).rejects.toThrow('Message not found or user not authorized.');
+    });
+  });
+
+  describe('deleteMessagesSince', () => {
+    it('should delete messages only for the authenticated user', async () => {
+      mockSchema.findOne().lean.mockResolvedValueOnce({ createdAt: new Date() });
+      mockFindQuery.deleteMany.mockResolvedValueOnce({ deletedCount: 1 });
+      const result = await deleteMessagesSince(mockReq, {
+        messageId: 'msg123',
+        conversationId: 'convo123',
+      });
+      expect(mockSchema.findOne).toHaveBeenCalledWith({ messageId: 'msg123', user: 'user123' });
+      expect(mockSchema.find).not.toHaveBeenCalled();
+      expect(result).toBeUndefined();
+    });
+
+    it('should return undefined if no message is found', async () => {
+      mockSchema.findOne().lean.mockResolvedValueOnce(null);
+      const result = await deleteMessagesSince(mockReq, {
+        messageId: 'nonexistent',
+        conversationId: 'convo123',
+      });
+      expect(result).toBeUndefined();
+    });
+  });
+
+  describe('getMessages', () => {
+    it('should retrieve messages with the correct filter', async () => {
+      const filter = { conversationId: 'convo123' };
+      await getMessages(filter);
+      expect(mockSchema.find).toHaveBeenCalledWith(filter);
+      expect(mockFindQuery.sort).toHaveBeenCalledWith({ createdAt: 1 });
+      expect(mockFindQuery.lean).toHaveBeenCalled();
+    });
+  });
+
+  describe('deleteMessages', () => {
+    it('should delete messages with the correct filter', async () => {
+      await deleteMessages({ user: 'user123' });
+      expect(mockSchema.deleteMany).toHaveBeenCalledWith({ user: 'user123' });
+    });
+  });
+
+  describe('Conversation Hijacking Prevention', () => {
+    it('should not allow editing a message in another user\'s conversation', async () => {
+      const attackerReq = { user: { id: 'attacker123' } };
+      const victimConversationId = 'victim-convo-123';
+      const victimMessageId = 'victim-msg-123';
+
+      mockSchema.findOneAndUpdate.mockResolvedValue(null);
+
+      await expect(
+        updateMessage(attackerReq, {
+          messageId: victimMessageId,
+          conversationId: victimConversationId,
+          text: 'Hacked message',
+        }),
+      ).rejects.toThrow('Message not found or user not authorized.');
+
+      expect(mockSchema.findOneAndUpdate).toHaveBeenCalledWith(
+        { messageId: victimMessageId, user: 'attacker123' },
+        expect.anything(),
+        expect.anything(),
+      );
+    });
+
+    it('should not allow deleting messages from another user\'s conversation', async () => {
+      const attackerReq = { user: { id: 'attacker123' } };
+      const victimConversationId = 'victim-convo-123';
+      const victimMessageId = 'victim-msg-123';
+
+      mockSchema.findOne().lean.mockResolvedValueOnce(null); // Simulating message not found for this user
+      const result = await deleteMessagesSince(attackerReq, {
+        messageId: victimMessageId,
+        conversationId: victimConversationId,
+      });
+
+      expect(result).toBeUndefined();
+      expect(mockSchema.findOne).toHaveBeenCalledWith({
+        messageId: victimMessageId,
+        user: 'attacker123',
+      });
+    });
+
+    it('should not allow inserting a new message into another user\'s conversation', async () => {
+      const attackerReq = { user: { id: 'attacker123' } };
+      const victimConversationId = uuidv4(); // Use a valid UUID
+
+      await expect(
+        saveMessage(attackerReq, {
+          conversationId: victimConversationId,
+          text: 'Inserted malicious message',
+          messageId: 'new-msg-123',
+        }),
+      ).resolves.not.toThrow(); // It should not throw an error
+
+      // Check that the message was saved with the attacker's user ID
+      expect(mockSchema.findOneAndUpdate).toHaveBeenCalledWith(
+        { messageId: 'new-msg-123', user: 'attacker123' },
+        expect.objectContaining({
+          user: 'attacker123',
+          conversationId: victimConversationId,
+        }),
+        expect.anything(),
+      );
+    });
+
+    it('should allow retrieving messages from any conversation', async () => {
+      const victimConversationId = 'victim-convo-123';
+
+      await getMessages({ conversationId: victimConversationId });
+
+      expect(mockSchema.find).toHaveBeenCalledWith({
+        conversationId: victimConversationId,
+      });
+
+      mockSchema.find.mockReturnValueOnce({
+        select: jest.fn().mockReturnThis(),
+        sort: jest.fn().mockReturnThis(),
+        lean: jest.fn().mockResolvedValue([{ text: 'Test message' }]),
+      });
+
+      const result = await getMessages({ conversationId: victimConversationId });
+      expect(result).toEqual([{ text: 'Test message' }]);
+    });
+  });
+});

api/models/Project.js

@@ -0,0 +1,90 @@
+const { model } = require('mongoose');
+const projectSchema = require('~/models/schema/projectSchema');
+
+const Project = model('Project', projectSchema);
+
+/**
+ * Retrieve a project by ID and convert the found project document to a plain object.
+ *
+ * @param {string} projectId - The ID of the project to find and return as a plain object.
+ * @param {string|string[]} [fieldsToSelect] - The fields to include or exclude in the returned document.
+ * @returns {Promise<MongoProject>} A plain object representing the project document, or `null` if no project is found.
+ */
+const getProjectById = async function (projectId, fieldsToSelect = null) {
+  const query = Project.findById(projectId);
+
+  if (fieldsToSelect) {
+    query.select(fieldsToSelect);
+  }
+
+  return await query.lean();
+};
+
+/**
+ * Retrieve a project by name and convert the found project document to a plain object.
+ * If the project with the given name doesn't exist and the name is "instance", create it and return the lean version.
+ *
+ * @param {string} projectName - The name of the project to find or create.
+ * @param {string|string[]} [fieldsToSelect] - The fields to include or exclude in the returned document.
+ * @returns {Promise<MongoProject>} A plain object representing the project document.
+ */
+const getProjectByName = async function (projectName, fieldsToSelect = null) {
+  const query = { name: projectName };
+  const update = { $setOnInsert: { name: projectName } };
+  const options = {
+    new: true,
+    upsert: projectName === 'instance',
+    lean: true,
+    select: fieldsToSelect,
+  };
+
+  return await Project.findOneAndUpdate(query, update, options);
+};
+
+/**
+ * Add an array of prompt group IDs to a project's promptGroupIds array, ensuring uniqueness.
+ *
+ * @param {string} projectId - The ID of the project to update.
+ * @param {string[]} promptGroupIds - The array of prompt group IDs to add to the project.
+ * @returns {Promise<MongoProject>} The updated project document.
+ */
+const addGroupIdsToProject = async function (projectId, promptGroupIds) {
+  return await Project.findByIdAndUpdate(
+    projectId,
+    { $addToSet: { promptGroupIds: { $each: promptGroupIds } } },
+    { new: true },
+  );
+};
+
+/**
+ * Remove an array of prompt group IDs from a project's promptGroupIds array.
+ *
+ * @param {string} projectId - The ID of the project to update.
+ * @param {string[]} promptGroupIds - The array of prompt group IDs to remove from the project.
+ * @returns {Promise<MongoProject>} The updated project document.
+ */
+const removeGroupIdsFromProject = async function (projectId, promptGroupIds) {
+  return await Project.findByIdAndUpdate(
+    projectId,
+    { $pull: { promptGroupIds: { $in: promptGroupIds } } },
+    { new: true },
+  );
+};
+
+/**
+ * Remove a prompt group ID from all projects.
+ *
+ * @param {string} promptGroupId - The ID of the prompt group to remove from projects.
+ * @returns {Promise<void>}
+ */
+const removeGroupFromAllProjects = async (promptGroupId) => {
+  await Project.updateMany({}, { $pull: { promptGroupIds: promptGroupId } });
+};
+
+module.exports = {
+  getProjectById,
+  getProjectByName,
+  addGroupIdsToProject,
+  removeGroupIdsFromProject,
+  removeGroupFromAllProjects,
+};

api/models/Prompt.js

@@ -1,52 +1,528 @@
-const mongoose = require('mongoose');
+const { ObjectId } = require('mongodb');
+const { SystemRoles, SystemCategories } = require('librechat-data-provider');
+const {
+  getProjectByName,
+  addGroupIdsToProject,
+  removeGroupIdsFromProject,
+  removeGroupFromAllProjects,
+} = require('./Project');
+const { Prompt, PromptGroup } = require('./schema/promptSchema');
 const { logger } = require('~/config');
 
-const promptSchema = mongoose.Schema(
-  {
-    title: {
-      type: String,
-      required: true,
+/**
+ * Create a pipeline for the aggregation to get prompt groups
+ * @param {Object} query
+ * @param {number} skip
+ * @param {number} limit
+ * @returns {[Object]} - The pipeline for the aggregation
+ */
+const createGroupPipeline = (query, skip, limit) => {
+  return [
+    { $match: query },
+    { $sort: { createdAt: -1 } },
+    { $skip: skip },
+    { $limit: limit },
+    {
+      $lookup: {
+        from: 'prompts',
+        localField: 'productionId',
+        foreignField: '_id',
+        as: 'productionPrompt',
+      },
     },
-    prompt: {
-      type: String,
-      required: true,
+    { $unwind: { path: '$productionPrompt', preserveNullAndEmptyArrays: true } },
+    {
+      $project: {
+        name: 1,
+        numberOfGenerations: 1,
+        oneliner: 1,
+        category: 1,
+        projectIds: 1,
+        productionId: 1,
+        author: 1,
+        authorName: 1,
+        createdAt: 1,
+        updatedAt: 1,
+        'productionPrompt.prompt': 1,
+        // 'productionPrompt._id': 1,
+        // 'productionPrompt.type': 1,
+      },
     },
-    category: {
-      type: String,
-    },
-  },
-  { timestamps: true },
-);
+  ];
+};
 
-const Prompt = mongoose.models.Prompt || mongoose.model('Prompt', promptSchema);
+/**
+ * Create a pipeline for the aggregation to get all prompt groups
+ * @param {Object} query
+ * @param {Partial<MongoPromptGroup>} $project
+ * @returns {[Object]} - The pipeline for the aggregation
+ */
+const createAllGroupsPipeline = (
+  query,
+  $project = {
+    name: 1,
+    oneliner: 1,
+    category: 1,
+    author: 1,
+    authorName: 1,
+    createdAt: 1,
+    updatedAt: 1,
+    command: 1,
+    'productionPrompt.prompt': 1,
+  },
+) => {
+  return [
+    { $match: query },
+    { $sort: { createdAt: -1 } },
+    {
+      $lookup: {
+        from: 'prompts',
+        localField: 'productionId',
+        foreignField: '_id',
+        as: 'productionPrompt',
+      },
+    },
+    { $unwind: { path: '$productionPrompt', preserveNullAndEmptyArrays: true } },
+    {
+      $project,
+    },
+  ];
+};
+
+/**
+ * Get all prompt groups with filters
+ * @param {Object} req
+ * @param {TPromptGroupsWithFilterRequest} filter
+ * @returns {Promise<PromptGroupListResponse>}
+ */
+const getAllPromptGroups = async (req, filter) => {
+  try {
+    const { name, ...query } = filter;
+
+    if (!query.author) {
+      throw new Error('Author is required');
+    }
+
+    let searchShared = true;
+    let searchSharedOnly = false;
+    if (name) {
+      query.name = new RegExp(name, 'i');
+    }
+    if (!query.category) {
+      delete query.category;
+    } else if (query.category === SystemCategories.MY_PROMPTS) {
+      searchShared = false;
+      delete query.category;
+    } else if (query.category === SystemCategories.NO_CATEGORY) {
+      query.category = '';
+    } else if (query.category === SystemCategories.SHARED_PROMPTS) {
+      searchSharedOnly = true;
+      delete query.category;
+    }
+
+    let combinedQuery = query;
+
+    if (searchShared) {
+      const project = await getProjectByName('instance', 'promptGroupIds');
+      if (project && project.promptGroupIds.length > 0) {
+        const projectQuery = { _id: { $in: project.promptGroupIds }, ...query };
+        delete projectQuery.author;
+        combinedQuery = searchSharedOnly ? projectQuery : { $or: [projectQuery, query] };
+      }
+    }
+
+    const promptGroupsPipeline = createAllGroupsPipeline(combinedQuery);
+    return await PromptGroup.aggregate(promptGroupsPipeline).exec();
+  } catch (error) {
+    console.error('Error getting all prompt groups', error);
+    return { message: 'Error getting all prompt groups' };
+  }
+};
+
+/**
+ * Get prompt groups with filters
+ * @param {Object} req
+ * @param {TPromptGroupsWithFilterRequest} filter
+ * @returns {Promise<PromptGroupListResponse>}
+ */
+const getPromptGroups = async (req, filter) => {
+  try {
+    const { pageNumber = 1, pageSize = 10, name, ...query } = filter;
+
+    const validatedPageNumber = Math.max(parseInt(pageNumber, 10), 1);
+    const validatedPageSize = Math.max(parseInt(pageSize, 10), 1);
+
+    if (!query.author) {
+      throw new Error('Author is required');
+    }
+
+    let searchShared = true;
+    let searchSharedOnly = false;
+    if (name) {
+      query.name = new RegExp(name, 'i');
+    }
+    if (!query.category) {
+      delete query.category;
+    } else if (query.category === SystemCategories.MY_PROMPTS) {
+      searchShared = false;
+      delete query.category;
+    } else if (query.category === SystemCategories.NO_CATEGORY) {
+      query.category = '';
+    } else if (query.category === SystemCategories.SHARED_PROMPTS) {
+      searchSharedOnly = true;
+      delete query.category;
+    }
+
+    let combinedQuery = query;
+
+    if (searchShared) {
+      // const projects = req.user.projects || []; // TODO: handle multiple projects
+      const project = await getProjectByName('instance', 'promptGroupIds');
+      if (project && project.promptGroupIds.length > 0) {
+        const projectQuery = { _id: { $in: project.promptGroupIds }, ...query };
+        delete projectQuery.author;
+        combinedQuery = searchSharedOnly ? projectQuery : { $or: [projectQuery, query] };
+      }
+    }
+
+    const skip = (validatedPageNumber - 1) * validatedPageSize;
+    const limit = validatedPageSize;
+
+    const promptGroupsPipeline = createGroupPipeline(combinedQuery, skip, limit);
+    const totalPromptGroupsPipeline = [{ $match: combinedQuery }, { $count: 'total' }];
+
+    const [promptGroupsResults, totalPromptGroupsResults] = await Promise.all([
+      PromptGroup.aggregate(promptGroupsPipeline).exec(),
+      PromptGroup.aggregate(totalPromptGroupsPipeline).exec(),
+    ]);
+
+    const promptGroups = promptGroupsResults;
+    const totalPromptGroups =
+      totalPromptGroupsResults.length > 0 ? totalPromptGroupsResults[0].total : 0;
+
+    return {
+      promptGroups,
+      pageNumber: validatedPageNumber.toString(),
+      pageSize: validatedPageSize.toString(),
+      pages: Math.ceil(totalPromptGroups / validatedPageSize).toString(),
+    };
+  } catch (error) {
+    console.error('Error getting prompt groups', error);
+    return { message: 'Error getting prompt groups' };
+  }
+};
 
 module.exports = {
-  savePrompt: async ({ title, prompt }) => {
+  getPromptGroups,
+  getAllPromptGroups,
+  /**
+   * Create a prompt and its respective group
+   * @param {TCreatePromptRecord} saveData
+   * @returns {Promise<TCreatePromptResponse>}
+   */
+  createPromptGroup: async (saveData) => {
     try {
-      await Prompt.create({
-        title,
-        prompt,
-      });
-      return { title, prompt };
+      const { prompt, group, author, authorName } = saveData;
+
+      let newPromptGroup = await PromptGroup.findOneAndUpdate(
+        { ...group, author, authorName, productionId: null },
+        { $setOnInsert: { ...group, author, authorName, productionId: null } },
+        { new: true, upsert: true },
+      )
+        .lean()
+        .select('-__v')
+        .exec();
+
+      const newPrompt = await Prompt.findOneAndUpdate(
+        { ...prompt, author, groupId: newPromptGroup._id },
+        { $setOnInsert: { ...prompt, author, groupId: newPromptGroup._id } },
+        { new: true, upsert: true },
+      )
+        .lean()
+        .select('-__v')
+        .exec();
+
+      newPromptGroup = await PromptGroup.findByIdAndUpdate(
+        newPromptGroup._id,
+        { productionId: newPrompt._id },
+        { new: true },
+      )
+        .lean()
+        .select('-__v')
+        .exec();
+
+      return {
+        prompt: newPrompt,
+        group: {
+          ...newPromptGroup,
+          productionPrompt: { prompt: newPrompt.prompt },
+        },
+      };
+    } catch (error) {
+      logger.error('Error saving prompt group', error);
+      throw new Error('Error saving prompt group');
+    }
+  },
+  /**
+   * Save a prompt
+   * @param {TCreatePromptRecord} saveData
+   * @returns {Promise<TCreatePromptResponse>}
+   */
+  savePrompt: async (saveData) => {
+    try {
+      const { prompt, author } = saveData;
+      const newPromptData = {
+        ...prompt,
+        author,
+      };
+
+      /** @type {TPrompt} */
+      let newPrompt;
+      try {
+        newPrompt = await Prompt.create(newPromptData);
+      } catch (error) {
+        if (error?.message?.includes('groupId_1_version_1')) {
+          await Prompt.db.collection('prompts').dropIndex('groupId_1_version_1');
+        } else {
+          throw error;
+        }
+        newPrompt = await Prompt.create(newPromptData);
+      }
+
+      return { prompt: newPrompt };
     } catch (error) {
       logger.error('Error saving prompt', error);
-      return { prompt: 'Error saving prompt' };
+      return { message: 'Error saving prompt' };
     }
   },
   getPrompts: async (filter) => {
     try {
-      return await Prompt.find(filter).lean();
+      return await Prompt.find(filter).sort({ createdAt: -1 }).lean();
     } catch (error) {
       logger.error('Error getting prompts', error);
-      return { prompt: 'Error getting prompts' };
+      return { message: 'Error getting prompts' };
     }
   },
-  deletePrompts: async (filter) => {
+  getPrompt: async (filter) => {
     try {
-      return await Prompt.deleteMany(filter);
+      if (filter.groupId) {
+        filter.groupId = new ObjectId(filter.groupId);
+      }
+      return await Prompt.findOne(filter).lean();
     } catch (error) {
-      logger.error('Error deleting prompts', error);
-      return { prompt: 'Error deleting prompts' };
+      logger.error('Error getting prompt', error);
+      return { message: 'Error getting prompt' };
+    }
+  },
+  /**
+   * Get prompt groups with filters
+   * @param {TGetRandomPromptsRequest} filter
+   * @returns {Promise<TGetRandomPromptsResponse>}
+   */
+  getRandomPromptGroups: async (filter) => {
+    try {
+      const result = await PromptGroup.aggregate([
+        {
+          $match: {
+            category: { $ne: '' },
+          },
+        },
+        {
+          $group: {
+            _id: '$category',
+            promptGroup: { $first: '$$ROOT' },
+          },
+        },
+        {
+          $replaceRoot: { newRoot: '$promptGroup' },
+        },
+        {
+          $sample: { size: +filter.limit + +filter.skip },
+        },
+        {
+          $skip: +filter.skip,
+        },
+        {
+          $limit: +filter.limit,
+        },
+      ]);
+      return { prompts: result };
+    } catch (error) {
+      logger.error('Error getting prompt groups', error);
+      return { message: 'Error getting prompt groups' };
+    }
+  },
+  getPromptGroupsWithPrompts: async (filter) => {
+    try {
+      return await PromptGroup.findOne(filter)
+        .populate({
+          path: 'prompts',
+          select: '-_id -__v -user',
+        })
+        .select('-_id -__v -user')
+        .lean();
+    } catch (error) {
+      logger.error('Error getting prompt groups', error);
+      return { message: 'Error getting prompt groups' };
+    }
+  },
+  getPromptGroup: async (filter) => {
+    try {
+      return await PromptGroup.findOne(filter).lean();
+    } catch (error) {
+      logger.error('Error getting prompt group', error);
+      return { message: 'Error getting prompt group' };
+    }
+  },
+  /**
+   * Deletes a prompt and its corresponding prompt group if it is the last prompt in the group.
+   *
+   * @param {Object} options - The options for deleting the prompt.
+   * @param {ObjectId|string} options.promptId - The ID of the prompt to delete.
+   * @param {ObjectId|string} options.groupId - The ID of the prompt's group.
+   * @param {ObjectId|string} options.author - The ID of the prompt's author.
+   * @param {string} options.role - The role of the prompt's author.
+   * @return {Promise<TDeletePromptResponse>} An object containing the result of the deletion.
+   * If the prompt was deleted successfully, the object will have a property 'prompt' with the value 'Prompt deleted successfully'.
+   * If the prompt group was deleted successfully, the object will have a property 'promptGroup' with the message 'Prompt group deleted successfully' and id of the deleted group.
+   * If there was an error deleting the prompt, the object will have a property 'message' with the value 'Error deleting prompt'.
+   */
+  deletePrompt: async ({ promptId, groupId, author, role }) => {
+    const query = { _id: promptId, groupId, author };
+    if (role === SystemRoles.ADMIN) {
+      delete query.author;
+    }
+    const { deletedCount } = await Prompt.deleteOne(query);
+    if (deletedCount === 0) {
+      throw new Error('Failed to delete the prompt');
+    }
+
+    const remainingPrompts = await Prompt.find({ groupId })
+      .select('_id')
+      .sort({ createdAt: 1 })
+      .lean();
+
+    if (remainingPrompts.length === 0) {
+      await PromptGroup.deleteOne({ _id: groupId });
+      await removeGroupFromAllProjects(groupId);
+
+      return {
+        prompt: 'Prompt deleted successfully',
+        promptGroup: {
+          message: 'Prompt group deleted successfully',
+          id: groupId,
+        },
+      };
+    } else {
+      const promptGroup = await PromptGroup.findById(groupId).lean();
+      if (promptGroup.productionId.toString() === promptId.toString()) {
+        await PromptGroup.updateOne(
+          { _id: groupId },
+          { productionId: remainingPrompts[remainingPrompts.length - 1]._id },
+        );
+      }
+
+      return { prompt: 'Prompt deleted successfully' };
+    }
+  },
+  /**
+   * Update prompt group
+   * @param {Partial<MongoPromptGroup>} filter - Filter to find prompt group
+   * @param {Partial<MongoPromptGroup>} data - Data to update
+   * @returns {Promise<TUpdatePromptGroupResponse>}
+   */
+  updatePromptGroup: async (filter, data) => {
+    try {
+      const updateOps = {};
+      if (data.removeProjectIds) {
+        for (const projectId of data.removeProjectIds) {
+          await removeGroupIdsFromProject(projectId, [filter._id]);
+        }
+
+        updateOps.$pull = { projectIds: { $in: data.removeProjectIds } };
+        delete data.removeProjectIds;
+      }
+
+      if (data.projectIds) {
+        for (const projectId of data.projectIds) {
+          await addGroupIdsToProject(projectId, [filter._id]);
+        }
+
+        updateOps.$addToSet = { projectIds: { $each: data.projectIds } };
+        delete data.projectIds;
+      }
+
+      const updateData = { ...data, ...updateOps };
+      const updatedDoc = await PromptGroup.findOneAndUpdate(filter, updateData, {
+        new: true,
+        upsert: false,
+      });
+
+      if (!updatedDoc) {
+        throw new Error('Prompt group not found');
+      }
+
+      return updatedDoc;
+    } catch (error) {
+      logger.error('Error updating prompt group', error);
+      return { message: 'Error updating prompt group' };
+    }
+  },
+  /**
+   * Function to make a prompt production based on its ID.
+   * @param {String} promptId - The ID of the prompt to make production.
+   * @returns {Object} The result of the production operation.
+   */
+  makePromptProduction: async (promptId) => {
+    try {
+      const prompt = await Prompt.findById(promptId).lean();
+
+      if (!prompt) {
+        throw new Error('Prompt not found');
+      }
+
+      await PromptGroup.findByIdAndUpdate(
+        prompt.groupId,
+        { productionId: prompt._id },
+        { new: true },
+      )
+        .lean()
+        .exec();
+
+      return {
+        message: 'Prompt production made successfully',
+      };
+    } catch (error) {
+      logger.error('Error making prompt production', error);
+      return { message: 'Error making prompt production' };
+    }
+  },
+  updatePromptLabels: async (_id, labels) => {
+    try {
+      const response = await Prompt.updateOne({ _id }, { $set: { labels } });
+      if (response.matchedCount === 0) {
+        return { message: 'Prompt not found' };
+      }
+      return { message: 'Prompt labels updated successfully' };
+    } catch (error) {
+      logger.error('Error updating prompt labels', error);
+      return { message: 'Error updating prompt labels' };
+    }
+  },
+  deletePromptGroup: async (_id) => {
+    try {
+      const response = await PromptGroup.deleteOne({ _id });
+
+      if (response.deletedCount === 0) {
+        return { promptGroup: 'Prompt group not found' };
+      }
+
+      await Prompt.deleteMany({ groupId: new ObjectId(_id) });
+      await removeGroupFromAllProjects(_id);
+      return { promptGroup: 'Prompt group deleted successfully' };
+    } catch (error) {
+      logger.error('Error deleting prompt group', error);
+      return { message: 'Error deleting prompt group' };
     }
   },
 };

api/models/Role.js

@@ -0,0 +1,153 @@
+const {
+  CacheKeys,
+  SystemRoles,
+  roleDefaults,
+  PermissionTypes,
+  removeNullishValues,
+  promptPermissionsSchema,
+  bookmarkPermissionsSchema,
+} = require('librechat-data-provider');
+const getLogStores = require('~/cache/getLogStores');
+const Role = require('~/models/schema/roleSchema');
+const { logger } = require('~/config');
+
+/**
+ * Retrieve a role by name and convert the found role document to a plain object.
+ * If the role with the given name doesn't exist and the name is a system defined role, create it and return the lean version.
+ *
+ * @param {string} roleName - The name of the role to find or create.
+ * @param {string|string[]} [fieldsToSelect] - The fields to include or exclude in the returned document.
+ * @returns {Promise<Object>} A plain object representing the role document.
+ */
+const getRoleByName = async function (roleName, fieldsToSelect = null) {
+  try {
+    const cache = getLogStores(CacheKeys.ROLES);
+    const cachedRole = await cache.get(roleName);
+    if (cachedRole) {
+      return cachedRole;
+    }
+    let query = Role.findOne({ name: roleName });
+    if (fieldsToSelect) {
+      query = query.select(fieldsToSelect);
+    }
+    let role = await query.lean().exec();
+
+    if (!role && SystemRoles[roleName]) {
+      role = roleDefaults[roleName];
+      role = await new Role(role).save();
+      await cache.set(roleName, role);
+      return role.toObject();
+    }
+    await cache.set(roleName, role);
+    return role;
+  } catch (error) {
+    throw new Error(`Failed to retrieve or create role: ${error.message}`);
+  }
+};
+
+/**
+ * Update role values by name.
+ *
+ * @param {string} roleName - The name of the role to update.
+ * @param {Partial<TRole>} updates - The fields to update.
+ * @returns {Promise<TRole>} Updated role document.
+ */
+const updateRoleByName = async function (roleName, updates) {
+  try {
+    const cache = getLogStores(CacheKeys.ROLES);
+    const role = await Role.findOneAndUpdate(
+      { name: roleName },
+      { $set: updates },
+      { new: true, lean: true },
+    )
+      .select('-__v')
+      .lean()
+      .exec();
+    await cache.set(roleName, role);
+    return role;
+  } catch (error) {
+    throw new Error(`Failed to update role: ${error.message}`);
+  }
+};
+
+const permissionSchemas = {
+  [PermissionTypes.PROMPTS]: promptPermissionsSchema,
+  [PermissionTypes.BOOKMARKS]: bookmarkPermissionsSchema,
+};
+
+/**
+ * Updates access permissions for a specific role and multiple permission types.
+ * @param {SystemRoles} roleName - The role to update.
+ * @param {Object.<PermissionTypes, Object.<Permissions, boolean>>} permissionsUpdate - Permissions to update and their values.
+ */
+async function updateAccessPermissions(roleName, permissionsUpdate) {
+  const updates = {};
+  for (const [permissionType, permissions] of Object.entries(permissionsUpdate)) {
+    if (permissionSchemas[permissionType]) {
+      updates[permissionType] = removeNullishValues(permissions);
+    }
+  }
+
+  if (Object.keys(updates).length === 0) {
+    return;
+  }
+
+  try {
+    const role = await getRoleByName(roleName);
+    if (!role) {
+      return;
+    }
+
+    const updatedPermissions = {};
+    let hasChanges = false;
+
+    for (const [permissionType, permissions] of Object.entries(updates)) {
+      const currentPermissions = role[permissionType] || {};
+      updatedPermissions[permissionType] = { ...currentPermissions };
+
+      for (const [permission, value] of Object.entries(permissions)) {
+        if (currentPermissions[permission] !== value) {
+          updatedPermissions[permissionType][permission] = value;
+          hasChanges = true;
+          logger.info(
+            `Updating '${roleName}' role ${permissionType} '${permission}' permission from ${currentPermissions[permission]} to: ${value}`,
+          );
+        }
+      }
+    }
+
+    if (hasChanges) {
+      await updateRoleByName(roleName, updatedPermissions);
+      logger.info(`Updated '${roleName}' role permissions`);
+    } else {
+      logger.info(`No changes needed for '${roleName}' role permissions`);
+    }
+  } catch (error) {
+    logger.error(`Failed to update ${roleName} role permissions:`, error);
+  }
+}
+
+/**
+ * Initialize default roles in the system.
+ * Creates the default roles (ADMIN, USER) if they don't exist in the database.
+ *
+ * @returns {Promise<void>}
+ */
+const initializeRoles = async function () {
+  const defaultRoles = [SystemRoles.ADMIN, SystemRoles.USER];
+
+  for (const roleName of defaultRoles) {
+    let role = await Role.findOne({ name: roleName }).select('name').lean();
+    if (!role) {
+      role = new Role(roleDefaults[roleName]);
+      await role.save();
+    }
+  }
+};
+
+module.exports = {
+  getRoleByName,
+  initializeRoles,
+  updateRoleByName,
+  updateAccessPermissions,
+};

api/models/Role.spec.js

@@ -0,0 +1,197 @@
+const mongoose = require('mongoose');
+const { MongoMemoryServer } = require('mongodb-memory-server');
+const { SystemRoles, PermissionTypes } = require('librechat-data-provider');
+const Role = require('~/models/schema/roleSchema');
+const { updateAccessPermissions } = require('~/models/Role');
+const getLogStores = require('~/cache/getLogStores');
+
+// Mock the cache
+jest.mock('~/cache/getLogStores', () => {
+  return jest.fn().mockReturnValue({
+    get: jest.fn(),
+    set: jest.fn(),
+    del: jest.fn(),
+  });
+});
+
+let mongoServer;
+
+beforeAll(async () => {
+  mongoServer = await MongoMemoryServer.create();
+  const mongoUri = mongoServer.getUri();
+  await mongoose.connect(mongoUri);
+});
+
+afterAll(async () => {
+  await mongoose.disconnect();
+  await mongoServer.stop();
+});
+
+beforeEach(async () => {
+  await Role.deleteMany({});
+  getLogStores.mockClear();
+});
+
+describe('updateAccessPermissions', () => {
+  it('should update permissions when changes are needed', async () => {
+    await new Role({
+      name: SystemRoles.USER,
+      [PermissionTypes.PROMPTS]: {
+        CREATE: true,
+        USE: true,
+        SHARED_GLOBAL: false,
+      },
+    }).save();
+
+    await updateAccessPermissions(SystemRoles.USER, {
+      [PermissionTypes.PROMPTS]: {
+        CREATE: true,
+        USE: true,
+        SHARED_GLOBAL: true,
+      },
+    });
+
+    const updatedRole = await Role.findOne({ name: SystemRoles.USER }).lean();
+    expect(updatedRole[PermissionTypes.PROMPTS]).toEqual({
+      CREATE: true,
+      USE: true,
+      SHARED_GLOBAL: true,
+    });
+  });
+
+  it('should not update permissions when no changes are needed', async () => {
+    await new Role({
+      name: SystemRoles.USER,
+      [PermissionTypes.PROMPTS]: {
+        CREATE: true,
+        USE: true,
+        SHARED_GLOBAL: false,
+      },
+    }).save();
+
+    await updateAccessPermissions(SystemRoles.USER, {
+      [PermissionTypes.PROMPTS]: {
+        CREATE: true,
+        USE: true,
+        SHARED_GLOBAL: false,
+      },
+    });
+
+    const updatedRole = await Role.findOne({ name: SystemRoles.USER }).lean();
+    expect(updatedRole[PermissionTypes.PROMPTS]).toEqual({
+      CREATE: true,
+      USE: true,
+      SHARED_GLOBAL: false,
+    });
+  });
+
+  it('should handle non-existent roles', async () => {
+    await updateAccessPermissions('NON_EXISTENT_ROLE', {
+      [PermissionTypes.PROMPTS]: {
+        CREATE: true,
+      },
+    });
+
+    const role = await Role.findOne({ name: 'NON_EXISTENT_ROLE' });
+    expect(role).toBeNull();
+  });
+
+  it('should update only specified permissions', async () => {
+    await new Role({
+      name: SystemRoles.USER,
+      [PermissionTypes.PROMPTS]: {
+        CREATE: true,
+        USE: true,
+        SHARED_GLOBAL: false,
+      },
+    }).save();
+
+    await updateAccessPermissions(SystemRoles.USER, {
+      [PermissionTypes.PROMPTS]: {
+        SHARED_GLOBAL: true,
+      },
+    });
+
+    const updatedRole = await Role.findOne({ name: SystemRoles.USER }).lean();
+    expect(updatedRole[PermissionTypes.PROMPTS]).toEqual({
+      CREATE: true,
+      USE: true,
+      SHARED_GLOBAL: true,
+    });
+  });
+
+  it('should handle partial updates', async () => {
+    await new Role({
+      name: SystemRoles.USER,
+      [PermissionTypes.PROMPTS]: {
+        CREATE: true,
+        USE: true,
+        SHARED_GLOBAL: false,
+      },
+    }).save();
+
+    await updateAccessPermissions(SystemRoles.USER, {
+      [PermissionTypes.PROMPTS]: {
+        USE: false,
+      },
+    });
+
+    const updatedRole = await Role.findOne({ name: SystemRoles.USER }).lean();
+    expect(updatedRole[PermissionTypes.PROMPTS]).toEqual({
+      CREATE: true,
+      USE: false,
+      SHARED_GLOBAL: false,
+    });
+  });
+
+  it('should update multiple permission types at once', async () => {
+    await new Role({
+      name: SystemRoles.USER,
+      [PermissionTypes.PROMPTS]: {
+        CREATE: true,
+        USE: true,
+        SHARED_GLOBAL: false,
+      },
+      [PermissionTypes.BOOKMARKS]: {
+        USE: true,
+      },
+    }).save();
+
+    await updateAccessPermissions(SystemRoles.USER, {
+      [PermissionTypes.PROMPTS]: { USE: false, SHARED_GLOBAL: true },
+      [PermissionTypes.BOOKMARKS]: { USE: false },
+    });
+
+    const updatedRole = await Role.findOne({ name: SystemRoles.USER }).lean();
+    expect(updatedRole[PermissionTypes.PROMPTS]).toEqual({
+      CREATE: true,
+      USE: false,
+      SHARED_GLOBAL: true,
+    });
+    expect(updatedRole[PermissionTypes.BOOKMARKS]).toEqual({
+      USE: false,
+    });
+  });
+
+  it('should handle updates for a single permission type', async () => {
+    await new Role({
+      name: SystemRoles.USER,
+      [PermissionTypes.PROMPTS]: {
+        CREATE: true,
+        USE: true,
+        SHARED_GLOBAL: false,
+      },
+    }).save();
+
+    await updateAccessPermissions(SystemRoles.USER, {
+      [PermissionTypes.PROMPTS]: { USE: false, SHARED_GLOBAL: true },
+    });
+
+    const updatedRole = await Role.findOne({ name: SystemRoles.USER }).lean();
+    expect(updatedRole[PermissionTypes.PROMPTS]).toEqual({
+      CREATE: true,
+      USE: false,
+      SHARED_GLOBAL: true,
+    });
+  });
+});

api/models/Session.js

@@ -1,6 +1,6 @@
-const crypto = require('crypto');
 const mongoose = require('mongoose');
 const signPayload = require('~/server/services/signPayload');
+const { hashToken } = require('~/server/utils/crypto');
 const { logger } = require('~/config');
 
 const { REFRESH_TOKEN_EXPIRY } = process.env ?? {};
@@ -39,8 +39,7 @@ sessionSchema.methods.generateRefreshToken = async function () {
       expirationTime: Math.floor((expiresIn - Date.now()) / 1000),
     });
 
-    const hash = crypto.createHash('sha256');
-    this.refreshTokenHash = hash.update(refreshToken).digest('hex');
+    this.refreshTokenHash = await hashToken(refreshToken);
 
     await this.save();
 

api/models/Share.js

@@ -1,89 +1,252 @@
-const crypto = require('crypto');
-const { getMessages } = require('./Message');
+const { nanoid } = require('nanoid');
+const { Constants } = require('librechat-data-provider');
 const SharedLink = require('./schema/shareSchema');
+const { getMessages } = require('./Message');
 const logger = require('~/config/winston');
 
-module.exports = {
-  SharedLink,
-  getSharedMessages: async (shareId) => {
-    try {
-      const share = await SharedLink.findOne({ shareId })
-        .populate({
-          path: 'messages',
-          select: '-_id -__v -user',
-        })
-        .select('-_id -__v -user')
-        .lean();
+/**
+ * Anonymizes a conversation ID
+ * @returns {string} The anonymized conversation ID
+ */
+function anonymizeConvoId() {
+  return `convo_${nanoid()}`;
+}
 
-      if (!share || !share.conversationId || !share.isPublic) {
-        return null;
-      }
+/**
+ * Anonymizes an assistant ID
+ * @returns {string} The anonymized assistant ID
+ */
+function anonymizeAssistantId() {
+  return `a_${nanoid()}`;
+}
 
-      return share;
-    } catch (error) {
-      logger.error('[getShare] Error getting share link', error);
-      return { message: 'Error getting share link' };
+/**
+ * Anonymizes a message ID
+ * @param {string} id - The original message ID
+ * @returns {string} The anonymized message ID
+ */
+function anonymizeMessageId(id) {
+  return id === Constants.NO_PARENT ? id : `msg_${nanoid()}`;
+}
+
+/**
+ * Anonymizes a conversation object
+ * @param {object} conversation - The conversation object
+ * @returns {object} The anonymized conversation object
+ */
+function anonymizeConvo(conversation) {
+  const newConvo = { ...conversation };
+  if (newConvo.assistant_id) {
+    newConvo.assistant_id = anonymizeAssistantId();
+  }
+  return newConvo;
+}
+
+/**
+ * Anonymizes messages in a conversation
+ * @param {TMessage[]} messages - The original messages
+ * @param {string} newConvoId - The new conversation ID
+ * @returns {TMessage[]} The anonymized messages
+ */
+function anonymizeMessages(messages, newConvoId) {
+  const idMap = new Map();
+  return messages.map((message) => {
+    const newMessageId = anonymizeMessageId(message.messageId);
+    idMap.set(message.messageId, newMessageId);
+
+    const anonymizedMessage = Object.assign(message, {
+      messageId: newMessageId,
+      parentMessageId:
+        idMap.get(message.parentMessageId) || anonymizeMessageId(message.parentMessageId),
+      conversationId: newConvoId,
+    });
+
+    if (anonymizedMessage.model && anonymizedMessage.model.startsWith('asst_')) {
+      anonymizedMessage.model = anonymizeAssistantId();
     }
-  },
 
-  getSharedLinks: async (user, pageNumber = 1, pageSize = 25, isPublic = true) => {
-    const query = { user, isPublic };
-    try {
-      const totalConvos = (await SharedLink.countDocuments(query)) || 1;
-      const totalPages = Math.ceil(totalConvos / pageSize);
-      const shares = await SharedLink.find(query)
+    return anonymizedMessage;
+  });
+}
+
+/**
+ * Retrieves shared messages for a given share ID
+ * @param {string} shareId - The share ID
+ * @returns {Promise<object|null>} The shared conversation data or null if not found
+ */
+async function getSharedMessages(shareId) {
+  try {
+    const share = await SharedLink.findOne({ shareId })
+      .populate({
+        path: 'messages',
+        select: '-_id -__v -user',
+      })
+      .select('-_id -__v -user')
+      .lean();
+
+    if (!share || !share.conversationId || !share.isPublic) {
+      return null;
+    }
+
+    const newConvoId = anonymizeConvoId();
+    return Object.assign(share, {
+      conversationId: newConvoId,
+      messages: anonymizeMessages(share.messages, newConvoId),
+    });
+  } catch (error) {
+    logger.error('[getShare] Error getting share link', error);
+    throw new Error('Error getting share link');
+  }
+}
+
+/**
+ * Retrieves shared links for a user
+ * @param {string} user - The user ID
+ * @param {number} [pageNumber=1] - The page number
+ * @param {number} [pageSize=25] - The page size
+ * @param {boolean} [isPublic=true] - Whether to retrieve public links only
+ * @returns {Promise<object>} The shared links and pagination data
+ */
+async function getSharedLinks(user, pageNumber = 1, pageSize = 25, isPublic = true) {
+  const query = { user, isPublic };
+  try {
+    const [totalConvos, sharedLinks] = await Promise.all([
+      SharedLink.countDocuments(query),
+      SharedLink.find(query)
         .sort({ updatedAt: -1 })
         .skip((pageNumber - 1) * pageSize)
         .limit(pageSize)
         .select('-_id -__v -user')
-        .lean();
+        .lean(),
+    ]);
 
-      return { sharedLinks: shares, pages: totalPages, pageNumber, pageSize };
-    } catch (error) {
-      logger.error('[getShareByPage] Error getting shares', error);
-      return { message: 'Error getting shares' };
-    }
-  },
+    const totalPages = Math.ceil((totalConvos || 1) / pageSize);
 
-  createSharedLink: async (user, { conversationId, ...shareData }) => {
+    return {
+      sharedLinks,
+      pages: totalPages,
+      pageNumber,
+      pageSize,
+    };
+  } catch (error) {
+    logger.error('[getShareByPage] Error getting shares', error);
+    throw new Error('Error getting shares');
+  }
+}
+
+/**
+ * Creates a new shared link
+ * @param {string} user - The user ID
+ * @param {object} shareData - The share data
+ * @param {string} shareData.conversationId - The conversation ID
+ * @returns {Promise<object>} The created shared link
+ */
+async function createSharedLink(user, { conversationId, ...shareData }) {
+  try {
     const share = await SharedLink.findOne({ conversationId }).select('-_id -__v -user').lean();
     if (share) {
-      return share;
-    }
-
-    try {
-      const shareId = crypto.randomUUID();
-      const messages = await getMessages({ conversationId });
-      const update = { ...shareData, shareId, messages, user };
-      return await SharedLink.findOneAndUpdate({ conversationId: conversationId, user }, update, {
-        new: true,
-        upsert: true,
+      const newConvoId = anonymizeConvoId();
+      const sharedConvo = anonymizeConvo(share);
+      return Object.assign(sharedConvo, {
+        conversationId: newConvoId,
+        messages: anonymizeMessages(share.messages, newConvoId),
       });
-    } catch (error) {
-      logger.error('[saveShareMessage] Error saving conversation', error);
-      return { message: 'Error saving conversation' };
     }
-  },
 
-  updateSharedLink: async (user, { conversationId, ...shareData }) => {
+    const shareId = nanoid();
+    const messages = await getMessages({ conversationId });
+    const update = { ...shareData, shareId, messages, user };
+    const newShare = await SharedLink.findOneAndUpdate({ conversationId, user }, update, {
+      new: true,
+      upsert: true,
+    }).lean();
+
+    const newConvoId = anonymizeConvoId();
+    const sharedConvo = anonymizeConvo(newShare);
+    return Object.assign(sharedConvo, {
+      conversationId: newConvoId,
+      messages: anonymizeMessages(newShare.messages, newConvoId),
+    });
+  } catch (error) {
+    logger.error('[createSharedLink] Error creating shared link', error);
+    throw new Error('Error creating shared link');
+  }
+}
+
+/**
+ * Updates an existing shared link
+ * @param {string} user - The user ID
+ * @param {object} shareData - The share data to update
+ * @param {string} shareData.conversationId - The conversation ID
+ * @returns {Promise<object>} The updated shared link
+ */
+async function updateSharedLink(user, { conversationId, ...shareData }) {
+  try {
     const share = await SharedLink.findOne({ conversationId }).select('-_id -__v -user').lean();
     if (!share) {
       return { message: 'Share not found' };
     }
-    // update messages to the latest
+
     const messages = await getMessages({ conversationId });
     const update = { ...shareData, messages, user };
-    return await SharedLink.findOneAndUpdate({ conversationId: conversationId, user }, update, {
+    const updatedShare = await SharedLink.findOneAndUpdate({ conversationId, user }, update, {
       new: true,
       upsert: false,
-    });
-  },
+    }).lean();
 
-  deleteSharedLink: async (user, { shareId }) => {
-    const share = await SharedLink.findOne({ shareId, user });
-    if (!share) {
-      return { message: 'Share not found' };
-    }
-    return await SharedLink.findOneAndDelete({ shareId, user });
-  },
+    const newConvoId = anonymizeConvoId();
+    const sharedConvo = anonymizeConvo(updatedShare);
+    return Object.assign(sharedConvo, {
+      conversationId: newConvoId,
+      messages: anonymizeMessages(updatedShare.messages, newConvoId),
+    });
+  } catch (error) {
+    logger.error('[updateSharedLink] Error updating shared link', error);
+    throw new Error('Error updating shared link');
+  }
+}
+
+/**
+ * Deletes a shared link
+ * @param {string} user - The user ID
+ * @param {object} params - The deletion parameters
+ * @param {string} params.shareId - The share ID to delete
+ * @returns {Promise<object>} The result of the deletion
+ */
+async function deleteSharedLink(user, { shareId }) {
+  try {
+    const result = await SharedLink.findOneAndDelete({ shareId, user });
+    return result ? { message: 'Share deleted successfully' } : { message: 'Share not found' };
+  } catch (error) {
+    logger.error('[deleteSharedLink] Error deleting shared link', error);
+    throw new Error('Error deleting shared link');
+  }
+}
+
+/**
+ * Deletes all shared links for a specific user
+ * @param {string} user - The user ID
+ * @returns {Promise<object>} The result of the deletion
+ */
+async function deleteAllSharedLinks(user) {
+  try {
+    const result = await SharedLink.deleteMany({ user });
+    return {
+      message: 'All shared links have been deleted successfully',
+      deletedCount: result.deletedCount,
+    };
+  } catch (error) {
+    logger.error('[deleteAllSharedLinks] Error deleting shared links', error);
+    throw new Error('Error deleting shared links');
+  }
+}
+
+module.exports = {
+  SharedLink,
+  getSharedLinks,
+  createSharedLink,
+  updateSharedLink,
+  deleteSharedLink,
+  getSharedMessages,
+  deleteAllSharedLinks,
 };

api/models/Token.js

@@ -0,0 +1,117 @@
+const tokenSchema = require('./schema/tokenSchema');
+const mongoose = require('mongoose');
+const { logger } = require('~/config');
+
+/**
+ * Token model.
+ * @type {mongoose.Model}
+ */
+const Token = mongoose.model('Token', tokenSchema);
+
+/**
+ * Creates a new Token instance.
+ * @param {Object} tokenData - The data for the new Token.
+ * @param {mongoose.Types.ObjectId} tokenData.userId - The user's ID. It is required.
+ * @param {String} tokenData.email - The user's email.
+ * @param {String} tokenData.token - The token. It is required.
+ * @param {Number} tokenData.expiresIn - The number of seconds until the token expires.
+ * @returns {Promise<mongoose.Document>} The new Token instance.
+ * @throws Will throw an error if token creation fails.
+ */
+async function createToken(tokenData) {
+  try {
+    const currentTime = new Date();
+    const expiresAt = new Date(currentTime.getTime() + tokenData.expiresIn * 1000);
+
+    const newTokenData = {
+      ...tokenData,
+      createdAt: currentTime,
+      expiresAt,
+    };
+
+    const newToken = new Token(newTokenData);
+    return await newToken.save();
+  } catch (error) {
+    logger.debug('An error occurred while creating token:', error);
+    throw error;
+  }
+}
+
+/**
+ * Finds a Token document that matches the provided query.
+ * @param {Object} query - The query to match against.
+ * @param {mongoose.Types.ObjectId|String} query.userId - The ID of the user.
+ * @param {String} query.token - The token value.
+ * @param {String} query.email - The email of the user.
+ * @returns {Promise<Object|null>} The matched Token document, or null if not found.
+ * @throws Will throw an error if the find operation fails.
+ */
+async function findToken(query) {
+  try {
+    const conditions = [];
+
+    if (query.userId) {
+      conditions.push({ userId: query.userId });
+    }
+    if (query.token) {
+      conditions.push({ token: query.token });
+    }
+    if (query.email) {
+      conditions.push({ email: query.email });
+    }
+
+    const token = await Token.findOne({
+      $and: conditions,
+    }).lean();
+
+    return token;
+  } catch (error) {
+    logger.debug('An error occurred while finding token:', error);
+    throw error;
+  }
+}
+
+/**
+ * Updates a Token document that matches the provided query.
+ * @param {Object} query - The query to match against.
+ * @param {mongoose.Types.ObjectId|String} query.userId - The ID of the user.
+ * @param {String} query.token - The token value.
+ * @param {Object} updateData - The data to update the Token with.
+ * @returns {Promise<mongoose.Document|null>} The updated Token document, or null if not found.
+ * @throws Will throw an error if the update operation fails.
+ */
+async function updateToken(query, updateData) {
+  try {
+    return await Token.findOneAndUpdate(query, updateData, { new: true });
+  } catch (error) {
+    logger.debug('An error occurred while updating token:', error);
+    throw error;
+  }
+}
+
+/**
+ * Deletes all Token documents that match the provided token, user ID, or email.
+ * @param {Object} query - The query to match against.
+ * @param {mongoose.Types.ObjectId|String} query.userId - The ID of the user.
+ * @param {String} query.token - The token value.
+ * @param {String} query.email - The email of the user.
+ * @returns {Promise<Object>} The result of the delete operation.
+ * @throws Will throw an error if the delete operation fails.
+ */
+async function deleteTokens(query) {
+  try {
+    return await Token.deleteMany({
+      $or: [{ userId: query.userId }, { token: query.token }, { email: query.email }],
+    });
+  } catch (error) {
+    logger.debug('An error occurred while deleting tokens:', error);
+    throw error;
+  }
+}
+
+module.exports = {
+  createToken,
+  findToken,
+  updateToken,
+  deleteTokens,
+};

api/models/Transaction.js

@@ -1,12 +1,12 @@
 const mongoose = require('mongoose');
-const { isEnabled } = require('../server/utils/handleText');
+const { isEnabled } = require('~/server/utils/handleText');
 const transactionSchema = require('./schema/transaction');
-const { getMultiplier } = require('./tx');
+const { getMultiplier, getCacheMultiplier } = require('./tx');
 const { logger } = require('~/config');
 const Balance = require('./Balance');
 const cancelRate = 1.15;
 
-// Method to calculate and set the tokenValue for a transaction
+/** Method to calculate and set the tokenValue for a transaction */
 transactionSchema.methods.calculateTokenValue = function () {
   if (!this.valueKey || !this.tokenType) {
     this.tokenValue = this.rawAmount;
@@ -21,15 +21,17 @@ transactionSchema.methods.calculateTokenValue = function () {
   }
 };
 
-// Static method to create a transaction and update the balance
-transactionSchema.statics.create = async function (transactionData) {
+/**
+ * Static method to create a transaction and update the balance
+ * @param {txData} txData - Transaction data.
+ */
+transactionSchema.statics.create = async function (txData) {
   const Transaction = this;
 
-  const transaction = new Transaction(transactionData);
-  transaction.endpointTokenConfig = transactionData.endpointTokenConfig;
+  const transaction = new Transaction(txData);
+  transaction.endpointTokenConfig = txData.endpointTokenConfig;
   transaction.calculateTokenValue();
 
-  // Save the transaction
   await transaction.save();
 
   if (!isEnabled(process.env.CHECK_BALANCE)) {
@@ -57,6 +59,109 @@ transactionSchema.statics.create = async function (transactionData) {
   };
 };
 
+/**
+ * Static method to create a structured transaction and update the balance
+ * @param {txData} txData - Transaction data.
+ */
+transactionSchema.statics.createStructured = async function (txData) {
+  const Transaction = this;
+
+  const transaction = new Transaction({
+    ...txData,
+    endpointTokenConfig: txData.endpointTokenConfig,
+  });
+
+  transaction.calculateStructuredTokenValue();
+
+  await transaction.save();
+
+  if (!isEnabled(process.env.CHECK_BALANCE)) {
+    return;
+  }
+
+  let balance = await Balance.findOne({ user: transaction.user }).lean();
+  let incrementValue = transaction.tokenValue;
+
+  if (balance && balance?.tokenCredits + incrementValue < 0) {
+    incrementValue = -balance.tokenCredits;
+  }
+
+  balance = await Balance.findOneAndUpdate(
+    { user: transaction.user },
+    { $inc: { tokenCredits: incrementValue } },
+    { upsert: true, new: true },
+  ).lean();
+
+  return {
+    rate: transaction.rate,
+    user: transaction.user.toString(),
+    balance: balance.tokenCredits,
+    [transaction.tokenType]: incrementValue,
+  };
+};
+
+/** Method to calculate token value for structured tokens */
+transactionSchema.methods.calculateStructuredTokenValue = function () {
+  if (!this.tokenType) {
+    this.tokenValue = this.rawAmount;
+    return;
+  }
+
+  const { model, endpointTokenConfig } = this;
+
+  if (this.tokenType === 'prompt') {
+    const inputMultiplier = getMultiplier({ tokenType: 'prompt', model, endpointTokenConfig });
+    const writeMultiplier =
+      getCacheMultiplier({ cacheType: 'write', model, endpointTokenConfig }) ?? inputMultiplier;
+    const readMultiplier =
+      getCacheMultiplier({ cacheType: 'read', model, endpointTokenConfig }) ?? inputMultiplier;
+
+    this.rateDetail = {
+      input: inputMultiplier,
+      write: writeMultiplier,
+      read: readMultiplier,
+    };
+
+    const totalPromptTokens =
+      Math.abs(this.inputTokens || 0) +
+      Math.abs(this.writeTokens || 0) +
+      Math.abs(this.readTokens || 0);
+
+    if (totalPromptTokens > 0) {
+      this.rate =
+        (Math.abs(inputMultiplier * (this.inputTokens || 0)) +
+          Math.abs(writeMultiplier * (this.writeTokens || 0)) +
+          Math.abs(readMultiplier * (this.readTokens || 0))) /
+        totalPromptTokens;
+    } else {
+      this.rate = Math.abs(inputMultiplier); // Default to input rate if no tokens
+    }
+
+    this.tokenValue = -(
+      Math.abs(this.inputTokens || 0) * inputMultiplier +
+      Math.abs(this.writeTokens || 0) * writeMultiplier +
+      Math.abs(this.readTokens || 0) * readMultiplier
+    );
+
+    this.rawAmount = -totalPromptTokens;
+  } else if (this.tokenType === 'completion') {
+    const multiplier = getMultiplier({ tokenType: this.tokenType, model, endpointTokenConfig });
+    this.rate = Math.abs(multiplier);
+    this.tokenValue = -Math.abs(this.rawAmount) * multiplier;
+    this.rawAmount = -Math.abs(this.rawAmount);
+  }
+
+  if (this.context && this.tokenType === 'completion' && this.context === 'incomplete') {
+    this.tokenValue = Math.ceil(this.tokenValue * cancelRate);
+    this.rate *= cancelRate;
+    if (this.rateDetail) {
+      this.rateDetail = Object.fromEntries(
+        Object.entries(this.rateDetail).map(([k, v]) => [k, v * cancelRate]),
+      );
+    }
+  }
+};
+
 const Transaction = mongoose.model('Transaction', transactionSchema);
 
 /**

api/models/Transaction.spec.js

@@ -0,0 +1,348 @@
+const mongoose = require('mongoose');
+const { MongoMemoryServer } = require('mongodb-memory-server');
+const Balance = require('./Balance');
+const { spendTokens, spendStructuredTokens } = require('./spendTokens');
+const { getMultiplier, getCacheMultiplier } = require('./tx');
+
+let mongoServer;
+
+beforeAll(async () => {
+  mongoServer = await MongoMemoryServer.create();
+  const mongoUri = mongoServer.getUri();
+  await mongoose.connect(mongoUri);
+});
+
+afterAll(async () => {
+  await mongoose.disconnect();
+  await mongoServer.stop();
+});
+
+beforeEach(async () => {
+  await mongoose.connection.dropDatabase();
+});
+
+describe('Regular Token Spending Tests', () => {
+  test('Balance should decrease when spending tokens with spendTokens', async () => {
+    // Arrange
+    const userId = new mongoose.Types.ObjectId();
+    const initialBalance = 10000000; // $10.00
+    await Balance.create({ user: userId, tokenCredits: initialBalance });
+
+    const model = 'gpt-3.5-turbo';
+    const txData = {
+      user: userId,
+      conversationId: 'test-conversation-id',
+      model,
+      context: 'test',
+      endpointTokenConfig: null,
+    };
+
+    const tokenUsage = {
+      promptTokens: 100,
+      completionTokens: 50,
+    };
+
+    // Act
+    process.env.CHECK_BALANCE = 'true';
+    await spendTokens(txData, tokenUsage);
+
+    // Assert
+    console.log('Initial Balance:', initialBalance);
+
+    const updatedBalance = await Balance.findOne({ user: userId });
+    console.log('Updated Balance:', updatedBalance.tokenCredits);
+
+    const promptMultiplier = getMultiplier({ model, tokenType: 'prompt' });
+    const completionMultiplier = getMultiplier({ model, tokenType: 'completion' });
+
+    const expectedPromptCost = tokenUsage.promptTokens * promptMultiplier;
+    const expectedCompletionCost = tokenUsage.completionTokens * completionMultiplier;
+    const expectedTotalCost = expectedPromptCost + expectedCompletionCost;
+    const expectedBalance = initialBalance - expectedTotalCost;
+
+    expect(updatedBalance.tokenCredits).toBeLessThan(initialBalance);
+    expect(updatedBalance.tokenCredits).toBeCloseTo(expectedBalance, 0);
+
+    console.log('Expected Total Cost:', expectedTotalCost);
+    console.log('Actual Balance Decrease:', initialBalance - updatedBalance.tokenCredits);
+  });
+
+  test('spendTokens should handle zero completion tokens', async () => {
+    // Arrange
+    const userId = new mongoose.Types.ObjectId();
+    const initialBalance = 10000000; // $10.00
+    await Balance.create({ user: userId, tokenCredits: initialBalance });
+
+    const model = 'gpt-3.5-turbo';
+    const txData = {
+      user: userId,
+      conversationId: 'test-conversation-id',
+      model,
+      context: 'test',
+      endpointTokenConfig: null,
+    };
+
+    const tokenUsage = {
+      promptTokens: 100,
+      completionTokens: 0,
+    };
+
+    // Act
+    process.env.CHECK_BALANCE = 'true';
+    await spendTokens(txData, tokenUsage);
+
+    // Assert
+    const updatedBalance = await Balance.findOne({ user: userId });
+
+    const promptMultiplier = getMultiplier({ model, tokenType: 'prompt' });
+    const expectedCost = tokenUsage.promptTokens * promptMultiplier;
+    expect(updatedBalance.tokenCredits).toBeCloseTo(initialBalance - expectedCost, 0);
+
+    console.log('Initial Balance:', initialBalance);
+    console.log('Updated Balance:', updatedBalance.tokenCredits);
+    console.log('Expected Cost:', expectedCost);
+  });
+
+  test('spendTokens should handle undefined token counts', async () => {
+    const userId = new mongoose.Types.ObjectId();
+    const initialBalance = 10000000; // $10.00
+    await Balance.create({ user: userId, tokenCredits: initialBalance });
+
+    const model = 'gpt-3.5-turbo';
+    const txData = {
+      user: userId,
+      conversationId: 'test-conversation-id',
+      model,
+      context: 'test',
+      endpointTokenConfig: null,
+    };
+
+    const tokenUsage = {};
+
+    const result = await spendTokens(txData, tokenUsage);
+
+    expect(result).toBeUndefined();
+  });
+
+  test('spendTokens should handle only prompt tokens', async () => {
+    const userId = new mongoose.Types.ObjectId();
+    const initialBalance = 10000000; // $10.00
+    await Balance.create({ user: userId, tokenCredits: initialBalance });
+
+    const model = 'gpt-3.5-turbo';
+    const txData = {
+      user: userId,
+      conversationId: 'test-conversation-id',
+      model,
+      context: 'test',
+      endpointTokenConfig: null,
+    };
+
+    const tokenUsage = { promptTokens: 100 };
+
+    await spendTokens(txData, tokenUsage);
+
+    const updatedBalance = await Balance.findOne({ user: userId });
+
+    const promptMultiplier = getMultiplier({ model, tokenType: 'prompt' });
+    const expectedCost = 100 * promptMultiplier;
+    expect(updatedBalance.tokenCredits).toBeCloseTo(initialBalance - expectedCost, 0);
+  });
+});
+
+describe('Structured Token Spending Tests', () => {
+  test('Balance should decrease and rawAmount should be set when spending a large number of structured tokens', async () => {
+    // Arrange
+    const userId = new mongoose.Types.ObjectId();
+    const initialBalance = 17613154.55; // $17.61
+    await Balance.create({ user: userId, tokenCredits: initialBalance });
+
+    const model = 'claude-3-5-sonnet';
+    const txData = {
+      user: userId,
+      conversationId: 'c23a18da-706c-470a-ac28-ec87ed065199',
+      model,
+      context: 'message',
+      endpointTokenConfig: null, // We'll use the default rates
+    };
+
+    const tokenUsage = {
+      promptTokens: {
+        input: 11,
+        write: 140522,
+        read: 0,
+      },
+      completionTokens: 5,
+    };
+
+    // Get the actual multipliers
+    const promptMultiplier = getMultiplier({ model, tokenType: 'prompt' });
+    const completionMultiplier = getMultiplier({ model, tokenType: 'completion' });
+    const writeMultiplier = getCacheMultiplier({ model, cacheType: 'write' });
+    const readMultiplier = getCacheMultiplier({ model, cacheType: 'read' });
+
+    console.log('Multipliers:', {
+      promptMultiplier,
+      completionMultiplier,
+      writeMultiplier,
+      readMultiplier,
+    });
+
+    // Act
+    process.env.CHECK_BALANCE = 'true';
+    const result = await spendStructuredTokens(txData, tokenUsage);
+
+    // Assert
+    console.log('Initial Balance:', initialBalance);
+    console.log('Updated Balance:', result.completion.balance);
+    console.log('Transaction Result:', result);
+
+    const expectedPromptCost =
+      tokenUsage.promptTokens.input * promptMultiplier +
+      tokenUsage.promptTokens.write * writeMultiplier +
+      tokenUsage.promptTokens.read * readMultiplier;
+    const expectedCompletionCost = tokenUsage.completionTokens * completionMultiplier;
+    const expectedTotalCost = expectedPromptCost + expectedCompletionCost;
+    const expectedBalance = initialBalance - expectedTotalCost;
+
+    console.log('Expected Cost:', expectedTotalCost);
+    console.log('Expected Balance:', expectedBalance);
+
+    expect(result.completion.balance).toBeLessThan(initialBalance);
+
+    // Allow for a small difference (e.g., 100 token credits, which is $0.0001)
+    const allowedDifference = 100;
+    expect(Math.abs(result.completion.balance - expectedBalance)).toBeLessThan(allowedDifference);
+
+    // Check if the decrease is approximately as expected
+    const balanceDecrease = initialBalance - result.completion.balance;
+    expect(balanceDecrease).toBeCloseTo(expectedTotalCost, 0);
+
+    // Check token values
+    const expectedPromptTokenValue = -(
+      tokenUsage.promptTokens.input * promptMultiplier +
+      tokenUsage.promptTokens.write * writeMultiplier +
+      tokenUsage.promptTokens.read * readMultiplier
+    );
+    const expectedCompletionTokenValue = -tokenUsage.completionTokens * completionMultiplier;
+
+    expect(result.prompt.prompt).toBeCloseTo(expectedPromptTokenValue, 1);
+    expect(result.completion.completion).toBe(expectedCompletionTokenValue);
+
+    console.log('Expected prompt tokenValue:', expectedPromptTokenValue);
+    console.log('Actual prompt tokenValue:', result.prompt.prompt);
+    console.log('Expected completion tokenValue:', expectedCompletionTokenValue);
+    console.log('Actual completion tokenValue:', result.completion.completion);
+  });
+
+  test('should handle zero completion tokens in structured spending', async () => {
+    const userId = new mongoose.Types.ObjectId();
+    const initialBalance = 17613154.55;
+    await Balance.create({ user: userId, tokenCredits: initialBalance });
+
+    const model = 'claude-3-5-sonnet';
+    const txData = {
+      user: userId,
+      conversationId: 'test-convo',
+      model,
+      context: 'message',
+    };
+
+    const tokenUsage = {
+      promptTokens: {
+        input: 10,
+        write: 100,
+        read: 5,
+      },
+      completionTokens: 0,
+    };
+
+    process.env.CHECK_BALANCE = 'true';
+    const result = await spendStructuredTokens(txData, tokenUsage);
+
+    expect(result.prompt).toBeDefined();
+    expect(result.completion).toBeUndefined();
+    expect(result.prompt.prompt).toBeLessThan(0);
+  });
+
+  test('should handle only prompt tokens in structured spending', async () => {
+    const userId = new mongoose.Types.ObjectId();
+    const initialBalance = 17613154.55;
+    await Balance.create({ user: userId, tokenCredits: initialBalance });
+
+    const model = 'claude-3-5-sonnet';
+    const txData = {
+      user: userId,
+      conversationId: 'test-convo',
+      model,
+      context: 'message',
+    };
+
+    const tokenUsage = {
+      promptTokens: {
+        input: 10,
+        write: 100,
+        read: 5,
+      },
+    };
+
+    process.env.CHECK_BALANCE = 'true';
+    const result = await spendStructuredTokens(txData, tokenUsage);
+
+    expect(result.prompt).toBeDefined();
+    expect(result.completion).toBeUndefined();
+    expect(result.prompt.prompt).toBeLessThan(0);
+  });
+
+  test('should handle undefined token counts in structured spending', async () => {
+    const userId = new mongoose.Types.ObjectId();
+    const initialBalance = 17613154.55;
+    await Balance.create({ user: userId, tokenCredits: initialBalance });
+
+    const model = 'claude-3-5-sonnet';
+    const txData = {
+      user: userId,
+      conversationId: 'test-convo',
+      model,
+      context: 'message',
+    };
+
+    const tokenUsage = {};
+
+    process.env.CHECK_BALANCE = 'true';
+    const result = await spendStructuredTokens(txData, tokenUsage);
+
+    expect(result).toEqual({
+      prompt: undefined,
+      completion: undefined,
+    });
+  });
+
+  test('should handle incomplete context for completion tokens', async () => {
+    const userId = new mongoose.Types.ObjectId();
+    const initialBalance = 17613154.55;
+    await Balance.create({ user: userId, tokenCredits: initialBalance });
+
+    const model = 'claude-3-5-sonnet';
+    const txData = {
+      user: userId,
+      conversationId: 'test-convo',
+      model,
+      context: 'incomplete',
+    };
+
+    const tokenUsage = {
+      promptTokens: {
+        input: 10,
+        write: 100,
+        read: 5,
+      },
+      completionTokens: 50,
+    };
+
+    process.env.CHECK_BALANCE = 'true';
+    const result = await spendStructuredTokens(txData, tokenUsage);
+
+    expect(result.completion.completion).toBeCloseTo(-50 * 15 * 1.15, 0); // Assuming multiplier is 15 and cancelRate is 1.15
+  });
+});

api/models/User.js

@@ -1,61 +1,5 @@
 const mongoose = require('mongoose');
-const bcrypt = require('bcryptjs');
-const signPayload = require('../server/services/signPayload');
-const userSchema = require('./schema/userSchema.js');
-const { SESSION_EXPIRY } = process.env ?? {};
-const expires = eval(SESSION_EXPIRY) ?? 1000 * 60 * 15;
-
-userSchema.methods.toJSON = function () {
-  return {
-    id: this._id,
-    provider: this.provider,
-    email: this.email,
-    name: this.name,
-    username: this.username,
-    avatar: this.avatar,
-    role: this.role,
-    emailVerified: this.emailVerified,
-    plugins: this.plugins,
-    createdAt: this.createdAt,
-    updatedAt: this.updatedAt,
-  };
-};
-
-userSchema.methods.generateToken = async function () {
-  return await signPayload({
-    payload: {
-      id: this._id,
-      username: this.username,
-      provider: this.provider,
-      email: this.email,
-    },
-    secret: process.env.JWT_SECRET,
-    expirationTime: expires / 1000,
-  });
-};
-
-userSchema.methods.comparePassword = function (candidatePassword, callback) {
-  bcrypt.compare(candidatePassword, this.password, (err, isMatch) => {
-    if (err) {
-      return callback(err);
-    }
-    callback(null, isMatch);
-  });
-};
-
-module.exports.hashPassword = async (password) => {
-  const hashedPassword = await new Promise((resolve, reject) => {
-    bcrypt.hash(password, 10, function (err, hash) {
-      if (err) {
-        reject(err);
-      } else {
-        resolve(hash);
-      }
-    });
-  });
-
-  return hashedPassword;
-};
+const userSchema = require('~/models/schema/userSchema');
 
 const User = mongoose.model('User', userSchema);
 

api/models/index.js

@@ -1,3 +1,22 @@
+const {
+  comparePassword,
+  deleteUserById,
+  generateToken,
+  getUserById,
+  updateUser,
+  createUser,
+  countUsers,
+  findUser,
+} = require('./userMethods');
+const {
+  findFileById,
+  createFile,
+  updateFile,
+  deleteFile,
+  deleteFiles,
+  getFiles,
+  updateFileUsage,
+} = require('./File');
 const {
   getMessages,
   saveMessage,
@@ -8,8 +27,22 @@ const {
 } = require('./Message');
 const { getConvoTitle, getConvo, saveConvo, deleteConvos } = require('./Conversation');
 const { getPreset, getPresets, savePreset, deletePresets } = require('./Preset');
-const { hashPassword, getUser, updateUser } = require('./userMethods');
-const {
+const { createToken, findToken, updateToken, deleteTokens } = require('./Token');
+const Session = require('./Session');
+const Balance = require('./Balance');
+const User = require('./User');
+const Key = require('./Key');
+
+module.exports = {
+  comparePassword,
+  deleteUserById,
+  generateToken,
+  getUserById,
+  updateUser,
+  createUser,
+  countUsers,
+  findUser,
+
   findFileById,
   createFile,
   updateFile,
@@ -17,21 +50,6 @@ const {
   deleteFiles,
   getFiles,
   updateFileUsage,
-} = require('./File');
-const Key = require('./Key');
-const User = require('./User');
-const Session = require('./Session');
-const Balance = require('./Balance');
-
-module.exports = {
-  User,
-  Key,
-  Session,
-  Balance,
-
-  hashPassword,
-  updateUser,
-  getUser,
 
   getMessages,
   saveMessage,
@@ -50,11 +68,13 @@ module.exports = {
   savePreset,
   deletePresets,
 
-  findFileById,
-  createFile,
-  updateFile,
-  deleteFile,
-  deleteFiles,
-  getFiles,
-  updateFileUsage,
+  createToken,
+  findToken,
+  updateToken,
+  deleteTokens,
+
+  User,
+  Key,
+  Session,
+  Balance,
 };

api/models/inviteUser.js

@@ -0,0 +1,70 @@
+const crypto = require('crypto');
+const bcrypt = require('bcryptjs');
+const mongoose = require('mongoose');
+const { createToken, findToken } = require('./Token');
+const logger = require('~/config/winston');
+
+/**
+ * @module inviteUser
+ * @description This module provides functions to create and get user invites
+ */
+
+/**
+ * @function createInvite
+ * @description This function creates a new user invite
+ * @param {string} email - The email of the user to invite
+ * @returns {Promise<Object>} A promise that resolves to the saved invite document
+ * @throws {Error} If there is an error creating the invite
+ */
+const createInvite = async (email) => {
+  try {
+    let token = crypto.randomBytes(32).toString('hex');
+    const hash = bcrypt.hashSync(token, 10);
+    const encodedToken = encodeURIComponent(token);
+
+    const fakeUserId = new mongoose.Types.ObjectId();
+
+    await createToken({
+      userId: fakeUserId,
+      email,
+      token: hash,
+      createdAt: Date.now(),
+      expiresIn: 604800,
+    });
+
+    return encodedToken;
+  } catch (error) {
+    logger.error('[createInvite] Error creating invite', error);
+    return { message: 'Error creating invite' };
+  }
+};
+
+/**
+ * @function getInvite
+ * @description This function retrieves a user invite
+ * @param {string} encodedToken - The token of the invite to retrieve
+ * @param {string} email - The email of the user to validate
+ * @returns {Promise<Object>} A promise that resolves to the retrieved invite document
+ * @throws {Error} If there is an error retrieving the invite, if the invite does not exist, or if the email does not match
+ */
+const getInvite = async (encodedToken, email) => {
+  try {
+    const token = decodeURIComponent(encodedToken);
+    const hash = bcrypt.hashSync(token, 10);
+    const invite = await findToken({ token: hash, email });
+
+    if (!invite) {
+      throw new Error('Invite not found or email does not match');
+    }
+
+    return invite;
+  } catch (error) {
+    logger.error('[getInvite] Error getting invite', error);
+    return { error: true, message: error.message };
+  }
+};
+
+module.exports = {
+  createInvite,
+  getInvite,
+};

api/models/plugins/mongoMeili.js

@@ -155,7 +155,7 @@ const createMeiliMongooseModel = function ({ index, attributesToIndex }) {
             function (results, value, key) {
               return { ...results, [key]: 1 };
             },
-            { _id: 1 },
+            { _id: 1, __v: 1 },
           ),
         ).lean();
 

api/models/schema/categories.js

@@ -0,0 +1,19 @@
+const mongoose = require('mongoose');
+const Schema = mongoose.Schema;
+
+const categoriesSchema = new Schema({
+  label: {
+    type: String,
+    required: true,
+    unique: true,
+  },
+  value: {
+    type: String,
+    required: true,
+    unique: true,
+  },
+});
+
+const categories = mongoose.model('categories', categoriesSchema);
+
+module.exports = { Categories: categories };

api/models/schema/conversationTagSchema.js

@@ -0,0 +1,31 @@
+const mongoose = require('mongoose');
+
+const conversationTagSchema = mongoose.Schema(
+  {
+    tag: {
+      type: String,
+      index: true,
+    },
+    user: {
+      type: String,
+      index: true,
+    },
+    description: {
+      type: String,
+      index: true,
+    },
+    count: {
+      type: Number,
+      default: 0,
+    },
+    position: {
+      type: Number,
+      default: 0,
+    },
+  },
+  { timestamps: true },
+);
+
+conversationTagSchema.index({ tag: 1, user: 1 }, { unique: true });
+
+module.exports = mongoose.model('ConversationTag', conversationTagSchema);

api/models/schema/convoSchema.js

@@ -42,6 +42,11 @@ const convoSchema = mongoose.Schema(
     invocationId: {
       type: Number,
     },
+    tags: {
+      type: [String],
+      default: [],
+      meiliIndex: true,
+    },
   },
   { timestamps: true },
 );
@@ -56,6 +61,7 @@ if (process.env.MEILI_HOST && process.env.MEILI_MASTER_KEY) {
 }
 
 convoSchema.index({ createdAt: 1, updatedAt: 1 });
+convoSchema.index({ conversationId: 1, user: 1 }, { unique: true });
 
 const Conversation = mongoose.models.Conversation || mongoose.model('Conversation', convoSchema);
 

api/models/schema/defaults.js

@@ -74,6 +74,10 @@ const conversationPreset = {
   resendImages: {
     type: Boolean,
   },
+  /* Anthropic only */
+  promptCache: {
+    type: Boolean,
+  },
   // files
   resendFiles: {
     type: Boolean,
@@ -103,6 +107,10 @@ const conversationPreset = {
   spec: {
     type: String,
   },
+  tags: {
+    type: [String],
+    default: [],
+  },
   tools: { type: [{ type: String }], default: undefined },
   maxContextTokens: {
     type: Number,

api/models/schema/fileSchema.js

@@ -3,9 +3,9 @@ const mongoose = require('mongoose');
 
 /**
  * @typedef {Object} MongoFile
- * @property {mongoose.Schema.Types.ObjectId} [_id] - MongoDB Document ID
+ * @property {ObjectId} [_id] - MongoDB Document ID
  * @property {number} [__v] - MongoDB Version Key
- * @property {mongoose.Schema.Types.ObjectId} user - User ID
+ * @property {ObjectId} user - User ID
  * @property {string} [conversationId] - Optional conversation ID
  * @property {string} file_id - File identifier
  * @property {string} [temp_file_id] - Temporary File identifier
@@ -14,17 +14,19 @@ const mongoose = require('mongoose');
  * @property {string} filepath - Location of the file
  * @property {'file'} object - Type of object, always 'file'
  * @property {string} type - Type of file
- * @property {number} usage - Number of uses of the file
+ * @property {number} [usage=0] - Number of uses of the file
  * @property {string} [context] - Context of the file origin
- * @property {boolean} [embedded] - Whether or not the file is embedded in vector db
+ * @property {boolean} [embedded=false] - Whether or not the file is embedded in vector db
  * @property {string} [model] - The model to identify the group region of the file (for Azure OpenAI hosting)
- * @property {string} [source] - The source of the file
+ * @property {string} [source] - The source of the file (e.g., from FileSources)
  * @property {number} [width] - Optional width of the file
  * @property {number} [height] - Optional height of the file
- * @property {Date} [expiresAt] - Optional height of the file
+ * @property {Date} [expiresAt] - Optional expiration date of the file
  * @property {Date} [createdAt] - Date when the file was created
  * @property {Date} [updatedAt] - Date when the file was updated
  */
+
+/** @type {MongooseSchema<MongoFile>} */
 const fileSchema = mongoose.Schema(
   {
     user: {
@@ -91,7 +93,7 @@ const fileSchema = mongoose.Schema(
     height: Number,
     expiresAt: {
       type: Date,
-      expires: 3600,
+      expires: 3600, // 1 hour in seconds
     },
   },
   {

api/models/schema/messageSchema.js

@@ -11,6 +11,7 @@ const messageSchema = mongoose.Schema(
     },
     conversationId: {
       type: String,
+      index: true,
       required: true,
       meiliIndex: true,
     },
@@ -128,7 +129,9 @@ if (process.env.MEILI_HOST && process.env.MEILI_MASTER_KEY) {
 }
 
 messageSchema.index({ createdAt: 1 });
+messageSchema.index({ messageId: 1, user: 1 }, { unique: true });
 
+/** @type {mongoose.Model<TMessage>} */
 const Message = mongoose.models.Message || mongoose.model('Message', messageSchema);
 
 module.exports = Message;

api/models/schema/projectSchema.js

@@ -0,0 +1,30 @@
+const { Schema } = require('mongoose');
+
+/**
+ * @typedef {Object} MongoProject
+ * @property {ObjectId} [_id] - MongoDB Document ID
+ * @property {string} name - The name of the project
+ * @property {ObjectId[]} promptGroupIds - Array of PromptGroup IDs associated with the project
+ * @property {Date} [createdAt] - Date when the project was created (added by timestamps)
+ * @property {Date} [updatedAt] - Date when the project was last updated (added by timestamps)
+ */
+
+const projectSchema = new Schema(
+  {
+    name: {
+      type: String,
+      required: true,
+      index: true,
+    },
+    promptGroupIds: {
+      type: [Schema.Types.ObjectId],
+      ref: 'PromptGroup',
+      default: [],
+    },
+  },
+  {
+    timestamps: true,
+  },
+);
+
+module.exports = projectSchema;

api/models/schema/promptSchema.js

@@ -0,0 +1,118 @@
+const mongoose = require('mongoose');
+const { Constants } = require('librechat-data-provider');
+const Schema = mongoose.Schema;
+
+/**
+ * @typedef {Object} MongoPromptGroup
+ * @property {ObjectId} [_id] - MongoDB Document ID
+ * @property {string} name - The name of the prompt group
+ * @property {ObjectId} author - The author of the prompt group
+ * @property {ObjectId} [projectId=null] - The project ID of the prompt group
+ * @property {ObjectId} [productionId=null] - The project ID of the prompt group
+ * @property {string} authorName - The name of the author of the prompt group
+ * @property {number} [numberOfGenerations=0] - Number of generations the prompt group has
+ * @property {string} [oneliner=''] - Oneliner description of the prompt group
+ * @property {string} [category=''] - Category of the prompt group
+ * @property {string} [command] - Command for the prompt group
+ * @property {Date} [createdAt] - Date when the prompt group was created (added by timestamps)
+ * @property {Date} [updatedAt] - Date when the prompt group was last updated (added by timestamps)
+ */
+
+const promptGroupSchema = new Schema(
+  {
+    name: {
+      type: String,
+      required: true,
+      index: true,
+    },
+    numberOfGenerations: {
+      type: Number,
+      default: 0,
+    },
+    oneliner: {
+      type: String,
+      default: '',
+    },
+    category: {
+      type: String,
+      default: '',
+      index: true,
+    },
+    projectIds: {
+      type: [Schema.Types.ObjectId],
+      ref: 'Project',
+      index: true,
+    },
+    productionId: {
+      type: Schema.Types.ObjectId,
+      ref: 'Prompt',
+      required: true,
+      index: true,
+    },
+    author: {
+      type: Schema.Types.ObjectId,
+      ref: 'User',
+      required: true,
+      index: true,
+    },
+    authorName: {
+      type: String,
+      required: true,
+    },
+    command: {
+      type: String,
+      index: true,
+      validate: {
+        validator: function (v) {
+          return v === undefined || v === null || v === '' || /^[a-z0-9-]+$/.test(v);
+        },
+        message: (props) =>
+          `${props.value} is not a valid command. Only lowercase alphanumeric characters and highfins (') are allowed.`,
+      },
+      maxlength: [
+        Constants.COMMANDS_MAX_LENGTH,
+        `Command cannot be longer than ${Constants.COMMANDS_MAX_LENGTH} characters`,
+      ],
+    },
+  },
+  {
+    timestamps: true,
+  },
+);
+
+const PromptGroup = mongoose.model('PromptGroup', promptGroupSchema);
+
+const promptSchema = new Schema(
+  {
+    groupId: {
+      type: Schema.Types.ObjectId,
+      ref: 'PromptGroup',
+      required: true,
+      index: true,
+    },
+    author: {
+      type: Schema.Types.ObjectId,
+      ref: 'User',
+      required: true,
+    },
+    prompt: {
+      type: String,
+      required: true,
+    },
+    type: {
+      type: String,
+      enum: ['text', 'chat'],
+      required: true,
+    },
+  },
+  {
+    timestamps: true,
+  },
+);
+
+const Prompt = mongoose.model('Prompt', promptSchema);
+
+promptSchema.index({ createdAt: 1, updatedAt: 1 });
+promptGroupSchema.index({ createdAt: 1, updatedAt: 1 });
+
+module.exports = { Prompt, PromptGroup };

api/models/schema/roleSchema.js

@@ -0,0 +1,35 @@
+const { PermissionTypes, Permissions } = require('librechat-data-provider');
+const mongoose = require('mongoose');
+
+const roleSchema = new mongoose.Schema({
+  name: {
+    type: String,
+    required: true,
+    unique: true,
+    index: true,
+  },
+  [PermissionTypes.BOOKMARKS]: {
+    [Permissions.USE]: {
+      type: Boolean,
+      default: true,
+    },
+  },
+  [PermissionTypes.PROMPTS]: {
+    [Permissions.SHARED_GLOBAL]: {
+      type: Boolean,
+      default: false,
+    },
+    [Permissions.USE]: {
+      type: Boolean,
+      default: true,
+    },
+    [Permissions.CREATE]: {
+      type: Boolean,
+      default: true,
+    },
+  },
+});
+
+const Role = mongoose.model('Role', roleSchema);
+
+module.exports = Role;

api/models/schema/tokenSchema.js

@@ -7,6 +7,9 @@ const tokenSchema = new Schema({
     required: true,
     ref: 'user',
   },
+  email: {
+    type: String,
+  },
   token: {
     type: String,
     required: true,
@@ -15,8 +18,13 @@ const tokenSchema = new Schema({
     type: Date,
     required: true,
     default: Date.now,
-    expires: 900,
+  },
+  expiresAt: {
+    type: Date,
+    required: true,
   },
 });
 
-module.exports = mongoose.model('Token', tokenSchema);
+tokenSchema.index({ expiresAt: 1 }, { expireAfterSeconds: 0 });
+
+module.exports = tokenSchema;

api/models/schema/transaction.js

@@ -30,6 +30,9 @@ const transactionSchema = mongoose.Schema(
     rate: Number,
     rawAmount: Number,
     tokenValue: Number,
+    inputTokens: { type: Number },
+    writeTokens: { type: Number },
+    readTokens: { type: Number },
   },
   {
     timestamps: true,

api/models/schema/userSchema.js

@@ -1,5 +1,36 @@
 const mongoose = require('mongoose');
+const { SystemRoles } = require('librechat-data-provider');
 
+/**
+ * @typedef {Object} MongoSession
+ * @property {string} [refreshToken] - The refresh token
+ */
+
+/**
+ * @typedef {Object} MongoUser
+ * @property {ObjectId} [_id] - MongoDB Document ID
+ * @property {string} [name] - The user's name
+ * @property {string} [username] - The user's username, in lowercase
+ * @property {string} email - The user's email address
+ * @property {boolean} emailVerified - Whether the user's email is verified
+ * @property {string} [password] - The user's password, trimmed with 8-128 characters
+ * @property {string} [avatar] - The URL of the user's avatar
+ * @property {string} provider - The provider of the user's account (e.g., 'local', 'google')
+ * @property {string} [role='USER'] - The role of the user
+ * @property {string} [googleId] - Optional Google ID for the user
+ * @property {string} [facebookId] - Optional Facebook ID for the user
+ * @property {string} [openidId] - Optional OpenID ID for the user
+ * @property {string} [ldapId] - Optional LDAP ID for the user
+ * @property {string} [githubId] - Optional GitHub ID for the user
+ * @property {string} [discordId] - Optional Discord ID for the user
+ * @property {Array} [plugins=[]] - List of plugins used by the user
+ * @property {Array.<MongoSession>} [refreshToken] - List of sessions with refresh tokens
+ * @property {Date} [expiresAt] - Optional expiration date of the file
+ * @property {Date} [createdAt] - Date when the user was created (added by timestamps)
+ * @property {Date} [updatedAt] - Date when the user was last updated (added by timestamps)
+ */
+
+/** @type {MongooseSchema<MongoSession>} */
 const Session = mongoose.Schema({
   refreshToken: {
     type: String,
@@ -7,6 +38,7 @@ const Session = mongoose.Schema({
   },
 });
 
+/** @type {MongooseSchema<MongoUser>} */
 const userSchema = mongoose.Schema(
   {
     name: {
@@ -47,7 +79,7 @@ const userSchema = mongoose.Schema(
     },
     role: {
       type: String,
-      default: 'USER',
+      default: SystemRoles.USER,
     },
     googleId: {
       type: String,
@@ -64,6 +96,11 @@ const userSchema = mongoose.Schema(
       unique: true,
       sparse: true,
     },
+    ldapId: {
+      type: String,
+      unique: true,
+      sparse: true,
+    },
     githubId: {
       type: String,
       unique: true,
@@ -81,6 +118,10 @@ const userSchema = mongoose.Schema(
     refreshToken: {
       type: [Session],
     },
+    expiresAt: {
+      type: Date,
+      expires: 604800, // 7 days in seconds
+    },
   },
   { timestamps: true },
 );

api/models/spendTokens.js

@@ -11,7 +11,7 @@ const { logger } = require('~/config');
  * @param {String} txData.conversationId - The ID of the conversation.
  * @param {String} txData.model - The model name.
  * @param {String} txData.context - The context in which the transaction is made.
- * @param {String} [txData.endpointTokenConfig] - The current endpoint token config.
+ * @param {EndpointTokenConfig} [txData.endpointTokenConfig] - The current endpoint token config.
  * @param {String} [txData.valueKey] - The value key (optional).
  * @param {Object} tokenUsage - The number of tokens used.
  * @param {Number} tokenUsage.promptTokens - The number of prompt tokens used.
@@ -32,38 +32,109 @@ const spendTokens = async (txData, tokenUsage) => {
   );
   let prompt, completion;
   try {
-    if (promptTokens >= 0) {
+    if (promptTokens !== undefined) {
       prompt = await Transaction.create({
         ...txData,
         tokenType: 'prompt',
-        rawAmount: -promptTokens,
+        rawAmount: -Math.max(promptTokens, 0),
       });
     }
 
-    if (!completionTokens) {
-      logger.debug('[spendTokens] !completionTokens', { prompt, completion });
-      return;
+    if (completionTokens !== undefined) {
+      completion = await Transaction.create({
+        ...txData,
+        tokenType: 'completion',
+        rawAmount: -Math.max(completionTokens, 0),
+      });
     }
 
-    completion = await Transaction.create({
-      ...txData,
-      tokenType: 'completion',
-      rawAmount: -completionTokens,
-    });
-
-    prompt &&
-      completion &&
+    if (prompt || completion) {
       logger.debug('[spendTokens] Transaction data record against balance:', {
         user: txData.user,
-        prompt: prompt.prompt,
-        promptRate: prompt.rate,
-        completion: completion.completion,
-        completionRate: completion.rate,
-        balance: completion.balance,
+        prompt: prompt?.prompt,
+        promptRate: prompt?.rate,
+        completion: completion?.completion,
+        completionRate: completion?.rate,
+        balance: completion?.balance ?? prompt?.balance,
       });
+    } else {
+      logger.debug('[spendTokens] No transactions incurred against balance');
+    }
   } catch (err) {
     logger.error('[spendTokens]', err);
   }
 };
 
-module.exports = spendTokens;
+/**
+ * Creates transactions to record the spending of structured tokens.
+ *
+ * @function
+ * @async
+ * @param {Object} txData - Transaction data.
+ * @param {mongoose.Schema.Types.ObjectId} txData.user - The user ID.
+ * @param {String} txData.conversationId - The ID of the conversation.
+ * @param {String} txData.model - The model name.
+ * @param {String} txData.context - The context in which the transaction is made.
+ * @param {EndpointTokenConfig} [txData.endpointTokenConfig] - The current endpoint token config.
+ * @param {String} [txData.valueKey] - The value key (optional).
+ * @param {Object} tokenUsage - The number of tokens used.
+ * @param {Object} tokenUsage.promptTokens - The number of prompt tokens used.
+ * @param {Number} tokenUsage.promptTokens.input - The number of input tokens.
+ * @param {Number} tokenUsage.promptTokens.write - The number of write tokens.
+ * @param {Number} tokenUsage.promptTokens.read - The number of read tokens.
+ * @param {Number} tokenUsage.completionTokens - The number of completion tokens used.
+ * @returns {Promise<void>} - Returns nothing.
+ * @throws {Error} - Throws an error if there's an issue creating the transactions.
+ */
+const spendStructuredTokens = async (txData, tokenUsage) => {
+  const { promptTokens, completionTokens } = tokenUsage;
+  logger.debug(
+    `[spendStructuredTokens] conversationId: ${txData.conversationId}${
+      txData?.context ? ` | Context: ${txData?.context}` : ''
+    } | Token usage: `,
+    {
+      promptTokens,
+      completionTokens,
+    },
+  );
+  let prompt, completion;
+  try {
+    if (promptTokens) {
+      const { input = 0, write = 0, read = 0 } = promptTokens;
+      prompt = await Transaction.createStructured({
+        ...txData,
+        tokenType: 'prompt',
+        inputTokens: -input,
+        writeTokens: -write,
+        readTokens: -read,
+      });
+    }
+
+    if (completionTokens) {
+      completion = await Transaction.create({
+        ...txData,
+        tokenType: 'completion',
+        rawAmount: -completionTokens,
+      });
+    }
+
+    if (prompt || completion) {
+      logger.debug('[spendStructuredTokens] Transaction data record against balance:', {
+        user: txData.user,
+        prompt: prompt?.prompt,
+        promptRate: prompt?.rate,
+        completion: completion?.completion,
+        completionRate: completion?.rate,
+        balance: completion?.balance ?? prompt?.balance,
+      });
+    } else {
+      logger.debug('[spendStructuredTokens] No transactions incurred against balance');
+    }
+  } catch (err) {
+    logger.error('[spendStructuredTokens]', err);
+  }
+
+  return { prompt, completion };
+};
+
+module.exports = { spendTokens, spendStructuredTokens };

api/models/spendTokens.spec.js

@@ -0,0 +1,197 @@
+const mongoose = require('mongoose');
+
+jest.mock('./Transaction', () => ({
+  Transaction: {
+    create: jest.fn(),
+    createStructured: jest.fn(),
+  },
+}));
+
+jest.mock('./Balance', () => ({
+  findOne: jest.fn(),
+  findOneAndUpdate: jest.fn(),
+}));
+
+jest.mock('~/config', () => ({
+  logger: {
+    debug: jest.fn(),
+    error: jest.fn(),
+  },
+}));
+
+// Import after mocking
+const { spendTokens, spendStructuredTokens } = require('./spendTokens');
+const { Transaction } = require('./Transaction');
+const Balance = require('./Balance');
+describe('spendTokens', () => {
+  beforeEach(() => {
+    jest.clearAllMocks();
+    process.env.CHECK_BALANCE = 'true';
+  });
+
+  it('should create transactions for both prompt and completion tokens', async () => {
+    const txData = {
+      user: new mongoose.Types.ObjectId(),
+      conversationId: 'test-convo',
+      model: 'gpt-3.5-turbo',
+      context: 'test',
+    };
+    const tokenUsage = {
+      promptTokens: 100,
+      completionTokens: 50,
+    };
+
+    Transaction.create.mockResolvedValueOnce({ tokenType: 'prompt', rawAmount: -100 });
+    Transaction.create.mockResolvedValueOnce({ tokenType: 'completion', rawAmount: -50 });
+    Balance.findOne.mockResolvedValue({ tokenCredits: 10000 });
+    Balance.findOneAndUpdate.mockResolvedValue({ tokenCredits: 9850 });
+
+    await spendTokens(txData, tokenUsage);
+
+    expect(Transaction.create).toHaveBeenCalledTimes(2);
+    expect(Transaction.create).toHaveBeenCalledWith(
+      expect.objectContaining({
+        tokenType: 'prompt',
+        rawAmount: -100,
+      }),
+    );
+    expect(Transaction.create).toHaveBeenCalledWith(
+      expect.objectContaining({
+        tokenType: 'completion',
+        rawAmount: -50,
+      }),
+    );
+  });
+
+  it('should handle zero completion tokens', async () => {
+    const txData = {
+      user: new mongoose.Types.ObjectId(),
+      conversationId: 'test-convo',
+      model: 'gpt-3.5-turbo',
+      context: 'test',
+    };
+    const tokenUsage = {
+      promptTokens: 100,
+      completionTokens: 0,
+    };
+
+    Transaction.create.mockResolvedValueOnce({ tokenType: 'prompt', rawAmount: -100 });
+    Transaction.create.mockResolvedValueOnce({ tokenType: 'completion', rawAmount: -0 });
+    Balance.findOne.mockResolvedValue({ tokenCredits: 10000 });
+    Balance.findOneAndUpdate.mockResolvedValue({ tokenCredits: 9850 });
+
+    await spendTokens(txData, tokenUsage);
+
+    expect(Transaction.create).toHaveBeenCalledTimes(2);
+    expect(Transaction.create).toHaveBeenCalledWith(
+      expect.objectContaining({
+        tokenType: 'prompt',
+        rawAmount: -100,
+      }),
+    );
+    expect(Transaction.create).toHaveBeenCalledWith(
+      expect.objectContaining({
+        tokenType: 'completion',
+        rawAmount: -0, // Changed from 0 to -0
+      }),
+    );
+  });
+
+  it('should handle undefined token counts', async () => {
+    const txData = {
+      user: new mongoose.Types.ObjectId(),
+      conversationId: 'test-convo',
+      model: 'gpt-3.5-turbo',
+      context: 'test',
+    };
+    const tokenUsage = {};
+
+    await spendTokens(txData, tokenUsage);
+
+    expect(Transaction.create).not.toHaveBeenCalled();
+  });
+
+  it('should not update balance when CHECK_BALANCE is false', async () => {
+    process.env.CHECK_BALANCE = 'false';
+    const txData = {
+      user: new mongoose.Types.ObjectId(),
+      conversationId: 'test-convo',
+      model: 'gpt-3.5-turbo',
+      context: 'test',
+    };
+    const tokenUsage = {
+      promptTokens: 100,
+      completionTokens: 50,
+    };
+
+    Transaction.create.mockResolvedValueOnce({ tokenType: 'prompt', rawAmount: -100 });
+    Transaction.create.mockResolvedValueOnce({ tokenType: 'completion', rawAmount: -50 });
+
+    await spendTokens(txData, tokenUsage);
+
+    expect(Transaction.create).toHaveBeenCalledTimes(2);
+    expect(Balance.findOne).not.toHaveBeenCalled();
+    expect(Balance.findOneAndUpdate).not.toHaveBeenCalled();
+  });
+
+  it('should create structured transactions for both prompt and completion tokens', async () => {
+    const txData = {
+      user: new mongoose.Types.ObjectId(),
+      conversationId: 'test-convo',
+      model: 'claude-3-5-sonnet',
+      context: 'test',
+    };
+    const tokenUsage = {
+      promptTokens: {
+        input: 10,
+        write: 100,
+        read: 5,
+      },
+      completionTokens: 50,
+    };
+
+    Transaction.createStructured.mockResolvedValueOnce({
+      rate: 3.75,
+      user: txData.user.toString(),
+      balance: 9570,
+      prompt: -430,
+    });
+    Transaction.create.mockResolvedValueOnce({
+      rate: 15,
+      user: txData.user.toString(),
+      balance: 8820,
+      completion: -750,
+    });
+
+    const result = await spendStructuredTokens(txData, tokenUsage);
+
+    expect(Transaction.createStructured).toHaveBeenCalledWith(
+      expect.objectContaining({
+        tokenType: 'prompt',
+        inputTokens: -10,
+        writeTokens: -100,
+        readTokens: -5,
+      }),
+    );
+    expect(Transaction.create).toHaveBeenCalledWith(
+      expect.objectContaining({
+        tokenType: 'completion',
+        rawAmount: -50,
+      }),
+    );
+    expect(result).toEqual({
+      prompt: expect.objectContaining({
+        rate: 3.75,
+        user: txData.user.toString(),
+        balance: 9570,
+        prompt: -430,
+      }),
+      completion: expect.objectContaining({
+        rate: 15,
+        user: txData.user.toString(),
+        balance: 8820,
+        completion: -750,
+      }),
+    });
+  });
+});

api/models/tx.js

@@ -1,35 +1,86 @@
 const { matchModelName } = require('../utils');
 const defaultRate = 6;
 
+/** AWS Bedrock pricing */
+const bedrockValues = {
+  'anthropic.claude-3-haiku-20240307-v1:0': { prompt: 0.25, completion: 1.25 },
+  'anthropic.claude-3-sonnet-20240229-v1:0': { prompt: 3.0, completion: 15.0 },
+  'anthropic.claude-3-opus-20240229-v1:0': { prompt: 15.0, completion: 75.0 },
+  'anthropic.claude-3-5-sonnet-20240620-v1:0': { prompt: 3.0, completion: 15.0 },
+  'anthropic.claude-v2:1': { prompt: 8.0, completion: 24.0 },
+  'anthropic.claude-instant-v1': { prompt: 0.8, completion: 2.4 },
+  'meta.llama2-13b-chat-v1': { prompt: 0.75, completion: 1.0 },
+  'meta.llama2-70b-chat-v1': { prompt: 1.95, completion: 2.56 },
+  'meta.llama3-8b-instruct-v1:0': { prompt: 0.3, completion: 0.6 },
+  'meta.llama3-70b-instruct-v1:0': { prompt: 2.65, completion: 3.5 },
+  'meta.llama3-1-8b-instruct-v1:0': { prompt: 0.3, completion: 0.6 },
+  'meta.llama3-1-70b-instruct-v1:0': { prompt: 2.65, completion: 3.5 },
+  'meta.llama3-1-405b-instruct-v1:0': { prompt: 5.32, completion: 16.0 },
+  'mistral.mistral-7b-instruct-v0:2': { prompt: 0.15, completion: 0.2 },
+  'mistral.mistral-small-2402-v1:0': { prompt: 0.15, completion: 0.2 },
+  'mistral.mixtral-8x7b-instruct-v0:1': { prompt: 0.45, completion: 0.7 },
+  'mistral.mistral-large-2402-v1:0': { prompt: 4.0, completion: 12.0 },
+  'mistral.mistral-large-2407-v1:0': { prompt: 3.0, completion: 9.0 },
+  'cohere.command-text-v14': { prompt: 1.5, completion: 2.0 },
+  'cohere.command-light-text-v14': { prompt: 0.3, completion: 0.6 },
+  'cohere.command-r-v1:0': { prompt: 0.5, completion: 1.5 },
+  'cohere.command-r-plus-v1:0': { prompt: 3.0, completion: 15.0 },
+  'ai21.j2-mid-v1': { prompt: 12.5, completion: 12.5 },
+  'ai21.j2-ultra-v1': { prompt: 18.8, completion: 18.8 },
+  'amazon.titan-text-lite-v1': { prompt: 0.15, completion: 0.2 },
+  'amazon.titan-text-express-v1': { prompt: 0.2, completion: 0.6 },
+};
+
+for (const [key, value] of Object.entries(bedrockValues)) {
+  bedrockValues[`bedrock/${key}`] = value;
+}
+
 /**
  * Mapping of model token sizes to their respective multipliers for prompt and completion.
  * The rates are 1 USD per 1M tokens.
  * @type {Object.<string, {prompt: number, completion: number}>}
  */
-const tokenValues = {
-  '8k': { prompt: 30, completion: 60 },
-  '32k': { prompt: 60, completion: 120 },
-  '4k': { prompt: 1.5, completion: 2 },
-  '16k': { prompt: 3, completion: 4 },
-  'gpt-3.5-turbo-1106': { prompt: 1, completion: 2 },
-  'gpt-4o': { prompt: 5, completion: 15 },
-  'gpt-4-1106': { prompt: 10, completion: 30 },
-  'gpt-3.5-turbo-0125': { prompt: 0.5, completion: 1.5 },
-  'claude-3-opus': { prompt: 15, completion: 75 },
-  'claude-3-sonnet': { prompt: 3, completion: 15 },
-  'claude-3-haiku': { prompt: 0.25, completion: 1.25 },
-  'claude-2.1': { prompt: 8, completion: 24 },
-  'claude-2': { prompt: 8, completion: 24 },
-  'claude-': { prompt: 0.8, completion: 2.4 },
-  'command-r-plus': { prompt: 3, completion: 15 },
-  'command-r': { prompt: 0.5, completion: 1.5 },
-  /* cohere doesn't have rates for the older command models,
+const tokenValues = Object.assign(
+  {
+    '8k': { prompt: 30, completion: 60 },
+    '32k': { prompt: 60, completion: 120 },
+    '4k': { prompt: 1.5, completion: 2 },
+    '16k': { prompt: 3, completion: 4 },
+    'gpt-3.5-turbo-1106': { prompt: 1, completion: 2 },
+    'gpt-4o-2024-08-06': { prompt: 2.5, completion: 10 },
+    'gpt-4o-mini': { prompt: 0.15, completion: 0.6 },
+    'gpt-4o': { prompt: 5, completion: 15 },
+    'gpt-4-1106': { prompt: 10, completion: 30 },
+    'gpt-3.5-turbo-0125': { prompt: 0.5, completion: 1.5 },
+    'claude-3-opus': { prompt: 15, completion: 75 },
+    'claude-3-sonnet': { prompt: 3, completion: 15 },
+    'claude-3-5-sonnet': { prompt: 3, completion: 15 },
+    'claude-3.5-sonnet': { prompt: 3, completion: 15 },
+    'claude-3-haiku': { prompt: 0.25, completion: 1.25 },
+    'claude-2.1': { prompt: 8, completion: 24 },
+    'claude-2': { prompt: 8, completion: 24 },
+    'claude-': { prompt: 0.8, completion: 2.4 },
+    'command-r-plus': { prompt: 3, completion: 15 },
+    'command-r': { prompt: 0.5, completion: 1.5 },
+    /* cohere doesn't have rates for the older command models,
   so this was from https://artificialanalysis.ai/models/command-light/providers */
-  command: { prompt: 0.38, completion: 0.38 },
-  // 'gemini-1.5': { prompt: 7, completion: 21 }, // May 2nd, 2024 pricing
-  // 'gemini': { prompt: 0.5, completion: 1.5 }, // May 2nd, 2024 pricing
-  'gemini-1.5': { prompt: 0, completion: 0 }, // currently free
-  gemini: { prompt: 0, completion: 0 }, // currently free
+    command: { prompt: 0.38, completion: 0.38 },
+    'gemini-1.5': { prompt: 7, completion: 21 }, // May 2nd, 2024 pricing
+    gemini: { prompt: 0.5, completion: 1.5 }, // May 2nd, 2024 pricing
+  },
+  bedrockValues,
+);
+
+/**
+ * Mapping of model token sizes to their respective multipliers for cached input, read and write.
+ * See Anthropic's documentation on this: https://docs.anthropic.com/en/docs/build-with-claude/prompt-caching#pricing
+ * The rates are 1 USD per 1M tokens.
+ * @type {Object.<string, {write: number, read: number }>}
+ */
+const cacheTokenValues = {
+  'claude-3.5-sonnet': { write: 3.75, read: 0.3 },
+  'claude-3-5-sonnet': { write: 3.75, read: 0.3 },
+  'claude-3-haiku': { write: 0.3, read: 0.03 },
 };
 
 /**
@@ -53,6 +104,10 @@ const getValueKey = (model, endpoint) => {
     return 'gpt-3.5-turbo-1106';
   } else if (modelName.includes('gpt-3.5')) {
     return '4k';
+  } else if (modelName.includes('gpt-4o-2024-08-06')) {
+    return 'gpt-4o-2024-08-06';
+  } else if (modelName.includes('gpt-4o-mini')) {
+    return 'gpt-4o-mini';
   } else if (modelName.includes('gpt-4o')) {
     return 'gpt-4o';
   } else if (modelName.includes('gpt-4-vision')) {
@@ -80,7 +135,7 @@ const getValueKey = (model, endpoint) => {
  *
  * @param {Object} params - The parameters for the function.
  * @param {string} [params.valueKey] - The key corresponding to the model name.
- * @param {string} [params.tokenType] - The type of token (e.g., 'prompt' or 'completion').
+ * @param {'prompt' | 'completion'} [params.tokenType] - The type of token (e.g., 'prompt' or 'completion').
  * @param {string} [params.model] - The model name to derive the value key from if not provided.
  * @param {string} [params.endpoint] - The endpoint name to derive the value key from if not provided.
  * @param {EndpointTokenConfig} [params.endpointTokenConfig] - The token configuration for the endpoint.
@@ -105,7 +160,41 @@ const getMultiplier = ({ valueKey, tokenType, model, endpoint, endpointTokenConf
   }
 
   // If we got this far, and values[tokenType] is undefined somehow, return a rough average of default multipliers
-  return tokenValues[valueKey][tokenType] ?? defaultRate;
+  return tokenValues[valueKey]?.[tokenType] ?? defaultRate;
 };
 
-module.exports = { tokenValues, getValueKey, getMultiplier, defaultRate };
+/**
+ * Retrieves the cache multiplier for a given value key and token type. If no value key is provided,
+ * it attempts to derive it from the model name.
+ *
+ * @param {Object} params - The parameters for the function.
+ * @param {string} [params.valueKey] - The key corresponding to the model name.
+ * @param {'write' | 'read'} [params.cacheType] - The type of token (e.g., 'write' or 'read').
+ * @param {string} [params.model] - The model name to derive the value key from if not provided.
+ * @param {string} [params.endpoint] - The endpoint name to derive the value key from if not provided.
+ * @param {EndpointTokenConfig} [params.endpointTokenConfig] - The token configuration for the endpoint.
+ * @returns {number | null} The multiplier for the given parameters, or `null` if not found.
+ */
+const getCacheMultiplier = ({ valueKey, cacheType, model, endpoint, endpointTokenConfig }) => {
+  if (endpointTokenConfig) {
+    return endpointTokenConfig?.[model]?.[cacheType] ?? null;
+  }
+
+  if (valueKey && cacheType) {
+    return cacheTokenValues[valueKey]?.[cacheType] ?? null;
+  }
+
+  if (!cacheType || !model) {
+    return null;
+  }
+
+  valueKey = getValueKey(model, endpoint);
+  if (!valueKey) {
+    return null;
+  }
+
+  // If we got this far, and values[cacheType] is undefined somehow, return a rough average of default multipliers
+  return cacheTokenValues[valueKey]?.[cacheType] ?? null;
+};
+
+module.exports = { tokenValues, getValueKey, getMultiplier, getCacheMultiplier, defaultRate };

api/models/tx.spec.js

@@ -1,4 +1,10 @@
-const { getValueKey, getMultiplier, defaultRate, tokenValues } = require('./tx');
+const {
+  defaultRate,
+  tokenValues,
+  getValueKey,
+  getMultiplier,
+  getCacheMultiplier,
+} = require('./tx');
 
 describe('getValueKey', () => {
   it('should return "16k" for model name containing "gpt-3.5-turbo-16k"', () => {
@@ -48,6 +54,41 @@ describe('getValueKey', () => {
     expect(getValueKey('gpt-4o-turbo')).toBe('gpt-4o');
     expect(getValueKey('gpt-4o-0125')).toBe('gpt-4o');
   });
+
+  it('should return "gpt-4o-mini" for model type of "gpt-4o-mini"', () => {
+    expect(getValueKey('gpt-4o-mini-2024-07-18')).toBe('gpt-4o-mini');
+    expect(getValueKey('openai/gpt-4o-mini')).toBe('gpt-4o-mini');
+    expect(getValueKey('gpt-4o-mini-0718')).toBe('gpt-4o-mini');
+    expect(getValueKey('gpt-4o-2024-08-06-0718')).not.toBe('gpt-4o');
+  });
+
+  it('should return "gpt-4o-2024-08-06" for model type of "gpt-4o-2024-08-06"', () => {
+    expect(getValueKey('gpt-4o-2024-08-06-2024-07-18')).toBe('gpt-4o-2024-08-06');
+    expect(getValueKey('openai/gpt-4o-2024-08-06')).toBe('gpt-4o-2024-08-06');
+    expect(getValueKey('gpt-4o-2024-08-06-0718')).toBe('gpt-4o-2024-08-06');
+    expect(getValueKey('gpt-4o-2024-08-06-0718')).not.toBe('gpt-4o');
+  });
+
+  it('should return "gpt-4o" for model type of "chatgpt-4o"', () => {
+    expect(getValueKey('chatgpt-4o-latest')).toBe('gpt-4o');
+    expect(getValueKey('openai/chatgpt-4o-latest')).toBe('gpt-4o');
+    expect(getValueKey('chatgpt-4o-latest-0916')).toBe('gpt-4o');
+    expect(getValueKey('chatgpt-4o-latest-0718')).toBe('gpt-4o');
+  });
+
+  it('should return "claude-3-5-sonnet" for model type of "claude-3-5-sonnet-"', () => {
+    expect(getValueKey('claude-3-5-sonnet-20240620')).toBe('claude-3-5-sonnet');
+    expect(getValueKey('anthropic/claude-3-5-sonnet')).toBe('claude-3-5-sonnet');
+    expect(getValueKey('claude-3-5-sonnet-turbo')).toBe('claude-3-5-sonnet');
+    expect(getValueKey('claude-3-5-sonnet-0125')).toBe('claude-3-5-sonnet');
+  });
+
+  it('should return "claude-3.5-sonnet" for model type of "claude-3.5-sonnet-"', () => {
+    expect(getValueKey('claude-3.5-sonnet-20240620')).toBe('claude-3.5-sonnet');
+    expect(getValueKey('anthropic/claude-3.5-sonnet')).toBe('claude-3.5-sonnet');
+    expect(getValueKey('claude-3.5-sonnet-turbo')).toBe('claude-3.5-sonnet');
+    expect(getValueKey('claude-3.5-sonnet-0125')).toBe('claude-3.5-sonnet');
+  });
 });
 
 describe('getMultiplier', () => {
@@ -102,6 +143,30 @@ describe('getMultiplier', () => {
     );
   });
 
+  it('should return the correct multiplier for gpt-4o-mini', () => {
+    const valueKey = getValueKey('gpt-4o-mini-2024-07-18');
+    expect(getMultiplier({ valueKey, tokenType: 'prompt' })).toBe(
+      tokenValues['gpt-4o-mini'].prompt,
+    );
+    expect(getMultiplier({ valueKey, tokenType: 'completion' })).toBe(
+      tokenValues['gpt-4o-mini'].completion,
+    );
+    expect(getMultiplier({ valueKey, tokenType: 'completion' })).not.toBe(
+      tokenValues['gpt-4-1106'].completion,
+    );
+  });
+
+  it('should return the correct multiplier for chatgpt-4o-latest', () => {
+    const valueKey = getValueKey('chatgpt-4o-latest');
+    expect(getMultiplier({ valueKey, tokenType: 'prompt' })).toBe(tokenValues['gpt-4o'].prompt);
+    expect(getMultiplier({ valueKey, tokenType: 'completion' })).toBe(
+      tokenValues['gpt-4o'].completion,
+    );
+    expect(getMultiplier({ valueKey, tokenType: 'completion' })).not.toBe(
+      tokenValues['gpt-4o-mini'].completion,
+    );
+  });
+
   it('should derive the valueKey from the model if not provided for new models', () => {
     expect(
       getMultiplier({ tokenType: 'prompt', model: 'gpt-3.5-turbo-1106-some-other-info' }),
@@ -126,3 +191,141 @@ describe('getMultiplier', () => {
     );
   });
 });
+
+describe('AWS Bedrock Model Tests', () => {
+  const awsModels = [
+    'anthropic.claude-3-haiku-20240307-v1:0',
+    'anthropic.claude-3-sonnet-20240229-v1:0',
+    'anthropic.claude-3-opus-20240229-v1:0',
+    'anthropic.claude-3-5-sonnet-20240620-v1:0',
+    'anthropic.claude-v2:1',
+    'anthropic.claude-instant-v1',
+    'meta.llama2-13b-chat-v1',
+    'meta.llama2-70b-chat-v1',
+    'meta.llama3-8b-instruct-v1:0',
+    'meta.llama3-70b-instruct-v1:0',
+    'meta.llama3-1-8b-instruct-v1:0',
+    'meta.llama3-1-70b-instruct-v1:0',
+    'meta.llama3-1-405b-instruct-v1:0',
+    'mistral.mistral-7b-instruct-v0:2',
+    'mistral.mistral-small-2402-v1:0',
+    'mistral.mixtral-8x7b-instruct-v0:1',
+    'mistral.mistral-large-2402-v1:0',
+    'mistral.mistral-large-2407-v1:0',
+    'cohere.command-text-v14',
+    'cohere.command-light-text-v14',
+    'cohere.command-r-v1:0',
+    'cohere.command-r-plus-v1:0',
+    'ai21.j2-mid-v1',
+    'ai21.j2-ultra-v1',
+    'amazon.titan-text-lite-v1',
+    'amazon.titan-text-express-v1',
+  ];
+
+  it('should return the correct prompt multipliers for all models', () => {
+    const results = awsModels.map((model) => {
+      const multiplier = getMultiplier({ valueKey: model, tokenType: 'prompt' });
+      return multiplier === tokenValues[model].prompt;
+    });
+    expect(results.every(Boolean)).toBe(true);
+  });
+
+  it('should return the correct completion multipliers for all models', () => {
+    const results = awsModels.map((model) => {
+      const multiplier = getMultiplier({ valueKey: model, tokenType: 'completion' });
+      return multiplier === tokenValues[model].completion;
+    });
+    expect(results.every(Boolean)).toBe(true);
+  });
+
+  it('should return the correct prompt multipliers for all models with Bedrock prefix', () => {
+    const results = awsModels.map((model) => {
+      const modelName = `bedrock/${model}`;
+      const multiplier = getMultiplier({ valueKey: modelName, tokenType: 'prompt' });
+      return multiplier === tokenValues[model].prompt;
+    });
+    expect(results.every(Boolean)).toBe(true);
+  });
+
+  it('should return the correct completion multipliers for all models with Bedrock prefix', () => {
+    const results = awsModels.map((model) => {
+      const modelName = `bedrock/${model}`;
+      const multiplier = getMultiplier({ valueKey: modelName, tokenType: 'completion' });
+      return multiplier === tokenValues[model].completion;
+    });
+    expect(results.every(Boolean)).toBe(true);
+  });
+});
+
+describe('getCacheMultiplier', () => {
+  it('should return the correct cache multiplier for a given valueKey and cacheType', () => {
+    expect(getCacheMultiplier({ valueKey: 'claude-3-5-sonnet', cacheType: 'write' })).toBe(3.75);
+    expect(getCacheMultiplier({ valueKey: 'claude-3-5-sonnet', cacheType: 'read' })).toBe(0.3);
+    expect(getCacheMultiplier({ valueKey: 'claude-3-haiku', cacheType: 'write' })).toBe(0.3);
+    expect(getCacheMultiplier({ valueKey: 'claude-3-haiku', cacheType: 'read' })).toBe(0.03);
+  });
+
+  it('should return null if cacheType is provided but not found in cacheTokenValues', () => {
+    expect(
+      getCacheMultiplier({ valueKey: 'claude-3-5-sonnet', cacheType: 'unknownType' }),
+    ).toBeNull();
+  });
+
+  it('should derive the valueKey from the model if not provided', () => {
+    expect(getCacheMultiplier({ cacheType: 'write', model: 'claude-3-5-sonnet-20240620' })).toBe(
+      3.75,
+    );
+    expect(getCacheMultiplier({ cacheType: 'read', model: 'claude-3-haiku-20240307' })).toBe(0.03);
+  });
+
+  it('should return null if only model or cacheType is missing', () => {
+    expect(getCacheMultiplier({ cacheType: 'write' })).toBeNull();
+    expect(getCacheMultiplier({ model: 'claude-3-5-sonnet' })).toBeNull();
+  });
+
+  it('should return null if derived valueKey does not match any known patterns', () => {
+    expect(getCacheMultiplier({ cacheType: 'write', model: 'gpt-4-some-other-info' })).toBeNull();
+  });
+
+  it('should handle endpointTokenConfig if provided', () => {
+    const endpointTokenConfig = {
+      'custom-model': {
+        write: 5,
+        read: 1,
+      },
+    };
+    expect(
+      getCacheMultiplier({ model: 'custom-model', cacheType: 'write', endpointTokenConfig }),
+    ).toBe(5);
+    expect(
+      getCacheMultiplier({ model: 'custom-model', cacheType: 'read', endpointTokenConfig }),
+    ).toBe(1);
+  });
+
+  it('should return null if model is not found in endpointTokenConfig', () => {
+    const endpointTokenConfig = {
+      'custom-model': {
+        write: 5,
+        read: 1,
+      },
+    };
+    expect(
+      getCacheMultiplier({ model: 'unknown-model', cacheType: 'write', endpointTokenConfig }),
+    ).toBeNull();
+  });
+
+  it('should handle models with "bedrock/" prefix', () => {
+    expect(
+      getCacheMultiplier({
+        model: 'bedrock/anthropic.claude-3-5-sonnet-20240620-v1:0',
+        cacheType: 'write',
+      }),
+    ).toBe(3.75);
+    expect(
+      getCacheMultiplier({
+        model: 'bedrock/anthropic.claude-3-haiku-20240307-v1:0',
+        cacheType: 'read',
+      }),
+    ).toBe(0.03);
+  });
+});

api/models/userMethods.js

@@ -1,28 +1,37 @@
 const bcrypt = require('bcryptjs');
+const signPayload = require('~/server/services/signPayload');
 const User = require('./User');
 
-const hashPassword = async (password) => {
-  const hashedPassword = await new Promise((resolve, reject) => {
-    bcrypt.hash(password, 10, function (err, hash) {
-      if (err) {
-        reject(err);
-      } else {
-        resolve(hash);
-      }
-    });
-  });
-
-  return hashedPassword;
-};
-
 /**
  * Retrieve a user by ID and convert the found user document to a plain object.
  *
  * @param {string} userId - The ID of the user to find and return as a plain object.
- * @returns {Promise<Object>} A plain object representing the user document, or `null` if no user is found.
+ * @param {string|string[]} [fieldsToSelect] - The fields to include or exclude in the returned document.
+ * @returns {Promise<MongoUser>} A plain object representing the user document, or `null` if no user is found.
  */
-const getUser = async function (userId) {
-  return await User.findById(userId).lean();
+const getUserById = async function (userId, fieldsToSelect = null) {
+  const query = User.findById(userId);
+
+  if (fieldsToSelect) {
+    query.select(fieldsToSelect);
+  }
+
+  return await query.lean();
+};
+
+/**
+ * Search for a single user based on partial data and return matching user document as plain object.
+ * @param {Partial<MongoUser>} searchCriteria - The partial data to use for searching the user.
+ * @param {string|string[]} [fieldsToSelect] - The fields to include or exclude in the returned document.
+ * @returns {Promise<MongoUser>} A plain object representing the user document, or `null` if no user is found.
+ */
+const findUser = async function (searchCriteria, fieldsToSelect = null) {
+  const query = User.findOne(searchCriteria);
+  if (fieldsToSelect) {
+    query.select(fieldsToSelect);
+  }
+
+  return await query.lean();
 };
 
 /**
@@ -30,17 +39,127 @@ const getUser = async function (userId) {
  *
  * @param {string} userId - The ID of the user to update.
  * @param {Object} updateData - An object containing the properties to update.
- * @returns {Promise<Object>} The updated user document as a plain object, or `null` if no user is found.
+ * @returns {Promise<MongoUser>} The updated user document as a plain object, or `null` if no user is found.
  */
 const updateUser = async function (userId, updateData) {
-  return await User.findByIdAndUpdate(userId, updateData, {
+  const updateOperation = {
+    $set: updateData,
+    $unset: { expiresAt: '' }, // Remove the expiresAt field to prevent TTL
+  };
+  return await User.findByIdAndUpdate(userId, updateOperation, {
     new: true,
     runValidators: true,
   }).lean();
 };
 
-module.exports = {
-  hashPassword,
-  updateUser,
-  getUser,
+/**
+ * Creates a new user, optionally with a TTL of 1 week.
+ * @param {MongoUser} data - The user data to be created, must contain user_id.
+ * @param {boolean} [disableTTL=true] - Whether to disable the TTL. Defaults to `true`.
+ * @param {boolean} [returnUser=false] - Whether to disable the TTL. Defaults to `true`.
+ * @returns {Promise<ObjectId>} A promise that resolves to the created user document ID.
+ * @throws {Error} If a user with the same user_id already exists.
+ */
+const createUser = async (data, disableTTL = true, returnUser = false) => {
+  const userData = {
+    ...data,
+    expiresAt: disableTTL ? null : new Date(Date.now() + 604800 * 1000), // 1 week in milliseconds
+  };
+
+  if (disableTTL) {
+    delete userData.expiresAt;
+  }
+
+  const user = await User.create(userData);
+  if (returnUser) {
+    return user.toObject();
+  }
+  return user._id;
+};
+
+/**
+ * Count the number of user documents in the collection based on the provided filter.
+ *
+ * @param {Object} [filter={}] - The filter to apply when counting the documents.
+ * @returns {Promise<number>} The count of documents that match the filter.
+ */
+const countUsers = async function (filter = {}) {
+  return await User.countDocuments(filter);
+};
+
+/**
+ * Delete a user by their unique ID.
+ *
+ * @param {string} userId - The ID of the user to delete.
+ * @returns {Promise<{ deletedCount: number }>} An object indicating the number of deleted documents.
+ */
+const deleteUserById = async function (userId) {
+  try {
+    const result = await User.deleteOne({ _id: userId });
+    if (result.deletedCount === 0) {
+      return { deletedCount: 0, message: 'No user found with that ID.' };
+    }
+    return { deletedCount: result.deletedCount, message: 'User was deleted successfully.' };
+  } catch (error) {
+    throw new Error('Error deleting user: ' + error.message);
+  }
+};
+
+const { SESSION_EXPIRY } = process.env ?? {};
+const expires = eval(SESSION_EXPIRY) ?? 1000 * 60 * 15;
+
+/**
+ * Generates a JWT token for a given user.
+ *
+ * @param {MongoUser} user - ID of the user for whom the token is being generated.
+ * @returns {Promise<string>} A promise that resolves to a JWT token.
+ */
+const generateToken = async (user) => {
+  if (!user) {
+    throw new Error('No user provided');
+  }
+
+  return await signPayload({
+    payload: {
+      id: user._id,
+      username: user.username,
+      provider: user.provider,
+      email: user.email,
+    },
+    secret: process.env.JWT_SECRET,
+    expirationTime: expires / 1000,
+  });
+};
+
+/**
+ * Compares the provided password with the user's password.
+ *
+ * @param {MongoUser} user - the user to compare password for.
+ * @param {string} candidatePassword - The password to test against the user's password.
+ * @returns {Promise<boolean>} A promise that resolves to a boolean indicating if the password matches.
+ */
+const comparePassword = async (user, candidatePassword) => {
+  if (!user) {
+    throw new Error('No user provided');
+  }
+
+  return new Promise((resolve, reject) => {
+    bcrypt.compare(candidatePassword, user.password, (err, isMatch) => {
+      if (err) {
+        reject(err);
+      }
+      resolve(isMatch);
+    });
+  });
+};
+
+module.exports = {
+  comparePassword,
+  deleteUserById,
+  generateToken,
+  getUserById,
+  countUsers,
+  createUser,
+  updateUser,
+  findUser,
 };

api/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@librechat/backend",
-  "version": "0.7.2",
+  "version": "v0.7.5-rc1",
   "description": "",
   "scripts": {
     "start": "echo 'please run this from the root directory'",
@@ -12,6 +12,7 @@
     "list-balances": "node ./list-balances.js",
     "user-stats": "node ./user-stats.js",
     "create-user": "node ./create-user.js",
+    "invite-user": "node ./invite-user.js",
     "ban-user": "node ./ban-user.js",
     "delete-user": "node ./delete-user.js"
   },
@@ -40,12 +41,12 @@
     "@keyv/redis": "^2.8.1",
     "@langchain/community": "^0.0.46",
     "@langchain/google-genai": "^0.0.11",
-    "@langchain/google-vertexai": "^0.0.5",
-    "agenda": "^5.0.0",
+    "@langchain/google-vertexai": "^0.0.17",
     "axios": "^1.3.4",
     "bcryptjs": "^2.4.3",
     "cheerio": "^1.0.0-rc.12",
     "cohere-ai": "^7.9.1",
+    "compression": "^1.7.4",
     "connect-redis": "^7.1.0",
     "cookie": "^0.5.0",
     "cors": "^2.8.5",
@@ -73,10 +74,11 @@
     "module-alias": "^2.2.3",
     "mongoose": "^7.1.1",
     "multer": "^1.4.5-lts.1",
+    "nanoid": "^3.3.7",
     "nodejs-gpt": "^1.37.4",
     "nodemailer": "^6.9.4",
     "ollama": "^0.5.0",
-    "openai": "4.36.0",
+    "openai": "^4.47.1",
     "openai-chat-tokens": "^0.2.8",
     "openid-client": "^5.4.2",
     "passport": "^0.6.0",
@@ -86,6 +88,7 @@
     "passport-github2": "^0.1.12",
     "passport-google-oauth20": "^2.0.0",
     "passport-jwt": "^4.0.1",
+    "passport-ldapauth": "^3.0.1",
     "passport-local": "^1.0.0",
     "pino": "^8.12.1",
     "sharp": "^0.32.6",
@@ -94,10 +97,12 @@
     "ua-parser-js": "^1.0.36",
     "winston": "^3.11.0",
     "winston-daily-rotate-file": "^4.7.1",
+    "ws": "^8.17.0",
     "zod": "^3.22.4"
   },
   "devDependencies": {
-    "jest": "^29.5.0",
+    "jest": "^29.7.0",
+    "mongodb-memory-server": "^10.0.0",
     "nodemon": "^3.0.1",
     "supertest": "^6.3.3"
   }

api/server/controllers/AskController.js

@@ -1,8 +1,9 @@
 const throttle = require('lodash/throttle');
-const { getResponseSender, Constants, EModelEndpoint } = require('librechat-data-provider');
+const { getResponseSender, Constants, CacheKeys, Time } = require('librechat-data-provider');
 const { createAbortController, handleAbortError } = require('~/server/middleware');
 const { sendMessage, createOnProgress } = require('~/server/utils');
-const { saveMessage, getConvo } = require('~/models');
+const { getLogStores } = require('~/cache');
+const { saveMessage } = require('~/models');
 const { logger } = require('~/config');
 
 const AskController = async (req, res, next, initializeClient, addTitle) => {
@@ -18,6 +19,7 @@ const AskController = async (req, res, next, initializeClient, addTitle) => {
   logger.debug('[AskController]', { text, conversationId, ...endpointOption });
 
   let userMessage;
+  let userMessagePromise;
   let promptTokens;
   let userMessageId;
   let responseMessageId;
@@ -34,6 +36,8 @@ const AskController = async (req, res, next, initializeClient, addTitle) => {
       if (key === 'userMessage') {
         userMessage = data[key];
         userMessageId = data[key].messageId;
+      } else if (key === 'userMessagePromise') {
+        userMessagePromise = data[key];
       } else if (key === 'responseMessageId') {
         responseMessageId = data[key];
       } else if (key === 'promptTokens') {
@@ -48,11 +52,13 @@ const AskController = async (req, res, next, initializeClient, addTitle) => {
 
   try {
     const { client } = await initializeClient({ req, res, endpointOption });
-    const unfinished = endpointOption.endpoint === EModelEndpoint.google ? false : true;
+    const messageCache = getLogStores(CacheKeys.MESSAGES);
     const { onProgress: progressCallback, getPartialText } = createOnProgress({
       onProgress: throttle(
         ({ text: partialText }) => {
-          saveMessage({
+          /*
+              const unfinished = endpointOption.endpoint === EModelEndpoint.google ? false : true;
+          messageCache.set(responseMessageId, {
             messageId: responseMessageId,
             sender,
             conversationId,
@@ -62,7 +68,10 @@ const AskController = async (req, res, next, initializeClient, addTitle) => {
             unfinished,
             error: false,
             user,
-          });
+          }, Time.FIVE_MINUTES);
+          */
+
+          messageCache.set(responseMessageId, partialText, Time.FIVE_MINUTES);
         },
         3000,
         { trailing: false },
@@ -74,6 +83,7 @@ const AskController = async (req, res, next, initializeClient, addTitle) => {
     const getAbortData = () => ({
       sender,
       conversationId,
+      userMessagePromise,
       messageId: responseMessageId,
       parentMessageId: overrideParentMessageId ?? userMessageId,
       text: getPartialText(),
@@ -81,7 +91,7 @@ const AskController = async (req, res, next, initializeClient, addTitle) => {
       promptTokens,
     });
 
-    const { abortController, onStart } = createAbortController(req, res, getAbortData);
+    const { abortController, onStart } = createAbortController(req, res, getAbortData, getReqData);
 
     res.on('close', () => {
       logger.debug('[AskController] Request closed');
@@ -105,11 +115,11 @@ const AskController = async (req, res, next, initializeClient, addTitle) => {
       getReqData,
       onStart,
       abortController,
-      onProgress: progressCallback.call(null, {
+      progressCallback,
+      progressOptions: {
         res,
-        text,
-        parentMessageId: overrideParentMessageId || userMessageId,
-      }),
+        // parentMessageId: overrideParentMessageId || userMessageId,
+      },
     };
 
     let response = await client.sendMessage(text, messageOptions);
@@ -120,7 +130,7 @@ const AskController = async (req, res, next, initializeClient, addTitle) => {
 
     response.endpoint = endpointOption.endpoint;
 
-    const conversation = await getConvo(user, conversationId);
+    const { conversation = {} } = await client.responsePromise;
     conversation.title =
       conversation && !conversation.title ? null : conversation?.title || 'New Chat';
 
@@ -140,10 +150,18 @@ const AskController = async (req, res, next, initializeClient, addTitle) => {
       });
       res.end();
 
-      await saveMessage({ ...response, user });
+      await saveMessage(
+        req,
+        { ...response, user },
+        { context: 'api/server/controllers/AskController.js - response end' },
+      );
     }
 
-    await saveMessage(userMessage);
+    if (!client.skipSaveUserMessage) {
+      await saveMessage(req, userMessage, {
+        context: 'api/server/controllers/AskController.js - don\'t skip saving user message',
+      });
+    }
 
     if (addTitle && parentMessageId === Constants.NO_PARENT && newConvo) {
       addTitle(req, {

api/server/controllers/AuthController.js

@@ -1,45 +1,29 @@
-const crypto = require('crypto');
 const cookies = require('cookie');
 const jwt = require('jsonwebtoken');
-const { Session, User } = require('~/models');
 const {
   registerUser,
   resetPassword,
   setAuthTokens,
   requestPasswordReset,
 } = require('~/server/services/AuthService');
+const { hashToken } = require('~/server/utils/crypto');
+const { Session, getUserById } = require('~/models');
 const { logger } = require('~/config');
 
 const registrationController = async (req, res) => {
   try {
     const response = await registerUser(req.body);
-    if (response.status === 200) {
-      const { status, user } = response;
-      let newUser = await User.findOne({ _id: user._id });
-      if (!newUser) {
-        newUser = new User(user);
-        await newUser.save();
-      }
-      const token = await setAuthTokens(user._id, res);
-      res.setHeader('Authorization', `Bearer ${token}`);
-      res.status(status).send({ user });
-    } else {
-      const { status, message } = response;
-      res.status(status).send({ message });
-    }
+    const { status, message } = response;
+    res.status(status).send({ message });
   } catch (err) {
     logger.error('[registrationController]', err);
     return res.status(500).json({ message: err.message });
   }
 };
 
-const getUserController = async (req, res) => {
-  return res.status(200).send(req.user);
-};
-
 const resetPasswordRequestController = async (req, res) => {
   try {
-    const resetService = await requestPasswordReset(req.body.email);
+    const resetService = await requestPasswordReset(req);
     if (resetService instanceof Error) {
       return res.status(400).json(resetService);
     } else {
@@ -77,7 +61,7 @@ const refreshController = async (req, res) => {
 
   try {
     const payload = jwt.verify(refreshToken, process.env.JWT_REFRESH_SECRET);
-    const user = await User.findOne({ _id: payload.id });
+    const user = await getUserById(payload.id, '-password -__v');
     if (!user) {
       return res.status(401).redirect('/login');
     }
@@ -86,20 +70,17 @@ const refreshController = async (req, res) => {
 
     if (process.env.NODE_ENV === 'CI') {
       const token = await setAuthTokens(userId, res);
-      const userObj = user.toJSON();
-      return res.status(200).send({ token, user: userObj });
+      return res.status(200).send({ token, user });
     }
 
     // Hash the refresh token
-    const hash = crypto.createHash('sha256');
-    const hashedToken = hash.update(refreshToken).digest('hex');
+    const hashedToken = await hashToken(refreshToken);
 
     // Find the session with the hashed refresh token
     const session = await Session.findOne({ user: userId, refreshTokenHash: hashedToken });
     if (session && session.expiration > new Date()) {
       const token = await setAuthTokens(userId, res, session._id);
-      const userObj = user.toJSON();
-      res.status(200).send({ token, user: userObj });
+      res.status(200).send({ token, user });
     } else if (req?.query?.retry) {
       // Retrying from a refresh token request that failed (401)
       res.status(403).send('No session found');
@@ -115,7 +96,6 @@ const refreshController = async (req, res) => {
 };
 
 module.exports = {
-  getUserController,
   refreshController,
   registrationController,
   resetPasswordController,

api/server/controllers/Balance.js

@@ -1,4 +1,4 @@
-const Balance = require('../../models/Balance');
+const Balance = require('~/models/Balance');
 
 async function balanceController(req, res) {
   const { tokenCredits: balance = '' } =

api/server/controllers/EditController.js

@@ -1,8 +1,9 @@
 const throttle = require('lodash/throttle');
-const { getResponseSender, EModelEndpoint } = require('librechat-data-provider');
+const { getResponseSender, CacheKeys, Time } = require('librechat-data-provider');
 const { createAbortController, handleAbortError } = require('~/server/middleware');
 const { sendMessage, createOnProgress } = require('~/server/utils');
-const { saveMessage, getConvo } = require('~/models');
+const { getLogStores } = require('~/cache');
+const { saveMessage } = require('~/models');
 const { logger } = require('~/config');
 
 const EditController = async (req, res, next, initializeClient) => {
@@ -27,6 +28,7 @@ const EditController = async (req, res, next, initializeClient) => {
   });
 
   let userMessage;
+  let userMessagePromise;
   let promptTokens;
   const sender = getResponseSender({
     ...endpointOption,
@@ -40,6 +42,8 @@ const EditController = async (req, res, next, initializeClient) => {
     for (let key in data) {
       if (key === 'userMessage') {
         userMessage = data[key];
+      } else if (key === 'userMessagePromise') {
+        userMessagePromise = data[key];
       } else if (key === 'responseMessageId') {
         responseMessageId = data[key];
       } else if (key === 'promptTokens') {
@@ -48,12 +52,14 @@ const EditController = async (req, res, next, initializeClient) => {
     }
   };
 
-  const unfinished = endpointOption.endpoint === EModelEndpoint.google ? false : true;
+  const messageCache = getLogStores(CacheKeys.MESSAGES);
   const { onProgress: progressCallback, getPartialText } = createOnProgress({
     generation,
     onProgress: throttle(
       ({ text: partialText }) => {
-        saveMessage({
+        /*
+          const unfinished = endpointOption.endpoint === EModelEndpoint.google ? false : true;
+        {
           messageId: responseMessageId,
           sender,
           conversationId,
@@ -64,7 +70,8 @@ const EditController = async (req, res, next, initializeClient) => {
           isEdited: true,
           error: false,
           user,
-        });
+        } */
+        messageCache.set(responseMessageId, partialText, Time.FIVE_MINUTES);
       },
       3000,
       { trailing: false },
@@ -73,6 +80,7 @@ const EditController = async (req, res, next, initializeClient) => {
 
   const getAbortData = () => ({
     conversationId,
+    userMessagePromise,
     messageId: responseMessageId,
     sender,
     parentMessageId: overrideParentMessageId ?? userMessageId,
@@ -81,7 +89,7 @@ const EditController = async (req, res, next, initializeClient) => {
     promptTokens,
   });
 
-  const { abortController, onStart } = createAbortController(req, res, getAbortData);
+  const { abortController, onStart } = createAbortController(req, res, getAbortData, getReqData);
 
   res.on('close', () => {
     logger.debug('[EditController] Request closed');
@@ -112,14 +120,14 @@ const EditController = async (req, res, next, initializeClient) => {
       getReqData,
       onStart,
       abortController,
-      onProgress: progressCallback.call(null, {
+      progressCallback,
+      progressOptions: {
         res,
-        text,
-        parentMessageId: overrideParentMessageId || userMessageId,
-      }),
+        // parentMessageId: overrideParentMessageId || userMessageId,
+      },
     });
 
-    const conversation = await getConvo(user, conversationId);
+    const { conversation = {} } = await client.responsePromise;
     conversation.title =
       conversation && !conversation.title ? null : conversation?.title || 'New Chat';
 
@@ -137,7 +145,11 @@ const EditController = async (req, res, next, initializeClient) => {
       });
       res.end();
 
-      await saveMessage({ ...response, user });
+      await saveMessage(
+        req,
+        { ...response, user },
+        { context: 'api/server/controllers/EditController.js - response end' },
+      );
     }
   } catch (error) {
     const partialText = getPartialText();

api/server/controllers/EndpointController.js

@@ -16,10 +16,28 @@ async function endpointController(req, res) {
   /** @type {TEndpointsConfig} */
   const mergedConfig = { ...defaultEndpointsConfig, ...customConfigEndpoints };
   if (mergedConfig[EModelEndpoint.assistants] && req.app.locals?.[EModelEndpoint.assistants]) {
-    const { disableBuilder, retrievalModels, capabilities, ..._rest } =
+    const { disableBuilder, retrievalModels, capabilities, version, ..._rest } =
       req.app.locals[EModelEndpoint.assistants];
+
     mergedConfig[EModelEndpoint.assistants] = {
       ...mergedConfig[EModelEndpoint.assistants],
+      version,
+      retrievalModels,
+      disableBuilder,
+      capabilities,
+    };
+  }
+
+  if (
+    mergedConfig[EModelEndpoint.azureAssistants] &&
+    req.app.locals?.[EModelEndpoint.azureAssistants]
+  ) {
+    const { disableBuilder, retrievalModels, capabilities, version, ..._rest } =
+      req.app.locals[EModelEndpoint.azureAssistants];
+
+    mergedConfig[EModelEndpoint.azureAssistants] = {
+      ...mergedConfig[EModelEndpoint.azureAssistants],
+      version,
       retrievalModels,
       disableBuilder,
       capabilities,

api/server/controllers/UserController.js

@@ -1,11 +1,37 @@
-const { updateUserPluginsService } = require('~/server/services/UserService');
+const {
+  Session,
+  Balance,
+  getFiles,
+  deleteFiles,
+  deleteConvos,
+  deletePresets,
+  deleteMessages,
+  deleteUserById,
+} = require('~/models');
 const { updateUserPluginAuth, deleteUserPluginAuth } = require('~/server/services/PluginService');
+const { updateUserPluginsService, deleteUserKey } = require('~/server/services/UserService');
+const { verifyEmail, resendVerificationEmail } = require('~/server/services/AuthService');
+const { processDeleteRequest } = require('~/server/services/Files/process');
+const { deleteAllSharedLinks } = require('~/models/Share');
+const { Transaction } = require('~/models/Transaction');
 const { logger } = require('~/config');
 
 const getUserController = async (req, res) => {
   res.status(200).send(req.user);
 };
 
+const deleteUserFiles = async (req) => {
+  try {
+    const userFiles = await getFiles({ user: req.user.id });
+    await processDeleteRequest({
+      req,
+      files: userFiles,
+    });
+  } catch (error) {
+    logger.error('[deleteUserFiles]', error);
+  }
+};
+
 const updateUserPluginsController = async (req, res) => {
   const { user } = req;
   const { pluginKey, action, auth, isAssistantTool } = req.body;
@@ -49,11 +75,68 @@ const updateUserPluginsController = async (req, res) => {
     res.status(200).send();
   } catch (err) {
     logger.error('[updateUserPluginsController]', err);
-    res.status(500).json({ message: err.message });
+    return res.status(500).json({ message: 'Something went wrong.' });
+  }
+};
+
+const deleteUserController = async (req, res) => {
+  const { user } = req;
+
+  try {
+    await deleteMessages({ user: user.id }); // delete user messages
+    await Session.deleteMany({ user: user.id }); // delete user sessions
+    await Transaction.deleteMany({ user: user.id }); // delete user transactions
+    await deleteUserKey({ userId: user.id, all: true }); // delete user keys
+    await Balance.deleteMany({ user: user._id }); // delete user balances
+    await deletePresets(user.id); // delete user presets
+    /* TODO: Delete Assistant Threads */
+    await deleteConvos(user.id); // delete user convos
+    await deleteUserPluginAuth(user.id, null, true); // delete user plugin auth
+    await deleteUserById(user.id); // delete user
+    await deleteAllSharedLinks(user.id); // delete user shared links
+    await deleteUserFiles(req); // delete user files
+    await deleteFiles(null, user.id); // delete database files in case of orphaned files from previous steps
+    /* TODO: queue job for cleaning actions and assistants of non-existant users */
+    logger.info(`User deleted account. Email: ${user.email} ID: ${user.id}`);
+    res.status(200).send({ message: 'User deleted' });
+  } catch (err) {
+    logger.error('[deleteUserController]', err);
+    return res.status(500).json({ message: 'Something went wrong.' });
+  }
+};
+
+const verifyEmailController = async (req, res) => {
+  try {
+    const verifyEmailService = await verifyEmail(req);
+    if (verifyEmailService instanceof Error) {
+      return res.status(400).json(verifyEmailService);
+    } else {
+      return res.status(200).json(verifyEmailService);
+    }
+  } catch (e) {
+    logger.error('[verifyEmailController]', e);
+    return res.status(500).json({ message: 'Something went wrong.' });
+  }
+};
+
+const resendVerificationController = async (req, res) => {
+  try {
+    const result = await resendVerificationEmail(req);
+    if (result instanceof Error) {
+      return res.status(400).json(result);
+    } else {
+      return res.status(200).json(result);
+    }
+  } catch (e) {
+    logger.error('[verifyEmailController]', e);
+    return res.status(500).json({ message: 'Something went wrong.' });
   }
 };
 
 module.exports = {
   getUserController,
+  deleteUserController,
+  verifyEmailController,
   updateUserPluginsController,
+  resendVerificationController,
 };

api/server/controllers/assistants/chatV1.js

@@ -1,14 +1,13 @@
 const { v4 } = require('uuid');
-const express = require('express');
 const {
   Constants,
   RunStatus,
   CacheKeys,
-  FileSources,
   ContentTypes,
   EModelEndpoint,
   ViolationTypes,
   ImageVisionTool,
+  checkOpenAIStorage,
   AssistantStreamEvents,
 } = require('librechat-data-provider');
 const {
@@ -21,44 +20,36 @@ const {
 } = require('~/server/services/Threads');
 const { sendResponse, sendMessage, sleep, isEnabled, countTokens } = require('~/server/utils');
 const { runAssistant, createOnTextProgress } = require('~/server/services/AssistantService');
-const { addTitle, initializeClient } = require('~/server/services/Endpoints/assistants');
+const validateAuthor = require('~/server/middleware/assistants/validateAuthor');
 const { formatMessage, createVisionPrompt } = require('~/app/clients/prompts');
 const { createRun, StreamRunManager } = require('~/server/services/Runs');
+const { addTitle } = require('~/server/services/Endpoints/assistants');
 const { getTransactions } = require('~/models/Transaction');
 const checkBalance = require('~/models/checkBalance');
 const { getConvo } = require('~/models/Conversation');
 const getLogStores = require('~/cache/getLogStores');
 const { getModelMaxTokens } = require('~/utils');
+const { getOpenAIClient } = require('./helpers');
 const { logger } = require('~/config');
 
-const router = express.Router();
-const {
-  setHeaders,
-  handleAbort,
-  validateModel,
-  handleAbortError,
-  // validateEndpoint,
-  buildEndpointOption,
-} = require('~/server/middleware');
-
-router.post('/abort', handleAbort());
-
 const ten_minutes = 1000 * 60 * 10;
 
 /**
  * @route POST /
  * @desc Chat with an assistant
  * @access Public
- * @param {express.Request} req - The request object, containing the request data.
- * @param {express.Response} res - The response object, used to send back a response.
+ * @param {object} req - The request object, containing the request data.
+ * @param {object} req.body - The request payload.
+ * @param {Express.Response} res - The response object, used to send back a response.
  * @returns {void}
  */
-router.post('/', validateModel, buildEndpointOption, setHeaders, async (req, res) => {
+const chatV1 = async (req, res) => {
   logger.debug('[/assistants/chat/] req.body', req.body);
 
   const {
     text,
     model,
+    endpoint,
     files = [],
     promptPrefix,
     assistant_id,
@@ -69,30 +60,6 @@ router.post('/', validateModel, buildEndpointOption, setHeaders, async (req, res
     parentMessageId: _parentId = Constants.NO_PARENT,
   } = req.body;
 
-  /** @type {Partial<TAssistantEndpoint>} */
-  const assistantsConfig = req.app.locals?.[EModelEndpoint.assistants];
-
-  if (assistantsConfig) {
-    const { supportedIds, excludedIds } = assistantsConfig;
-    const error = { message: 'Assistant not supported' };
-    if (supportedIds?.length && !supportedIds.includes(assistant_id)) {
-      return await handleAbortError(res, req, error, {
-        sender: 'System',
-        conversationId: convoId,
-        messageId: v4(),
-        parentMessageId: _messageId,
-        error,
-      });
-    } else if (excludedIds?.length && excludedIds.includes(assistant_id)) {
-      return await handleAbortError(res, req, error, {
-        sender: 'System',
-        conversationId: convoId,
-        messageId: v4(),
-        parentMessageId: _messageId,
-      });
-    }
-  }
-
   /** @type {OpenAIClient} */
   let openai;
   /** @type {string|undefined} - the current thread id */
@@ -138,7 +105,7 @@ router.post('/', validateModel, buildEndpointOption, setHeaders, async (req, res
       user: req.user.id,
       shouldSaveMessage: false,
       messageId: responseMessageId,
-      endpoint: EModelEndpoint.assistants,
+      endpoint,
     };
 
     if (error.message === 'Run cancelled') {
@@ -149,25 +116,26 @@ router.post('/', validateModel, buildEndpointOption, setHeaders, async (req, res
       logger.debug('[/assistants/chat/] Request aborted on close');
     } else if (/Files.*are invalid/.test(error.message)) {
       const errorMessage = `Files are invalid, or may not have uploaded yet.${
-        req.app.locals?.[EModelEndpoint.azureOpenAI].assistants
+        endpoint === EModelEndpoint.azureAssistants
           ? ' If using Azure OpenAI, files are only available in the region of the assistant\'s model at the time of upload.'
           : ''
       }`;
-      return sendResponse(res, messageData, errorMessage);
+      return sendResponse(req, res, messageData, errorMessage);
     } else if (error?.message?.includes('string too long')) {
       return sendResponse(
+        req,
         res,
         messageData,
         'Message too long. The Assistants API has a limit of 32,768 characters per message. Please shorten it and try again.',
       );
     } else if (error?.message?.includes(ViolationTypes.TOKEN_BALANCE)) {
-      return sendResponse(res, messageData, error.message);
+      return sendResponse(req, res, messageData, error.message);
     } else {
       logger.error('[/assistants/chat/]', error);
     }
 
     if (!openai || !thread_id || !run_id) {
-      return sendResponse(res, messageData, defaultErrorMessage);
+      return sendResponse(req, res, messageData, defaultErrorMessage);
     }
 
     await sleep(2000);
@@ -205,6 +173,7 @@ router.post('/', validateModel, buildEndpointOption, setHeaders, async (req, res
       const runMessages = await checkMessageGaps({
         openai,
         run_id,
+        endpoint,
         thread_id,
         conversationId,
         latestMessageId: responseMessageId,
@@ -253,10 +222,10 @@ router.post('/', validateModel, buildEndpointOption, setHeaders, async (req, res
       };
     } catch (error) {
       logger.error('[/assistants/chat/] Error finalizing error process', error);
-      return sendResponse(res, messageData, 'The Assistant run failed');
+      return sendResponse(req, res, messageData, 'The Assistant run failed');
     }
 
-    return sendResponse(res, finalEvent);
+    return sendResponse(req, res, finalEvent);
   };
 
   try {
@@ -311,8 +280,7 @@ router.post('/', validateModel, buildEndpointOption, setHeaders, async (req, res
       });
     };
 
-    /** @type {{ openai: OpenAIClient }} */
-    const { openai: _openai, client } = await initializeClient({
+    const { openai: _openai, client } = await getOpenAIClient({
       req,
       res,
       endpointOption: req.body.endpointOption,
@@ -320,6 +288,7 @@ router.post('/', validateModel, buildEndpointOption, setHeaders, async (req, res
     });
 
     openai = _openai;
+    await validateAuthor({ req, openai });
 
     if (previousMessages.length) {
       parentMessageId = previousMessages[previousMessages.length - 1].messageId;
@@ -370,10 +339,7 @@ router.post('/', validateModel, buildEndpointOption, setHeaders, async (req, res
 
       /** @type {MongoFile[]} */
       const attachments = await req.body.endpointOption.attachments;
-      if (
-        attachments &&
-        attachments.every((attachment) => attachment.source === FileSources.openai)
-      ) {
+      if (attachments && attachments.every((attachment) => checkOpenAIStorage(attachment.source))) {
         return;
       }
 
@@ -417,6 +383,9 @@ router.post('/', validateModel, buildEndpointOption, setHeaders, async (req, res
       return files;
     };
 
+    /** @type {Promise<Run>|undefined} */
+    let userMessagePromise;
+
     const initializeThread = async () => {
       /** @type {[ undefined | MongoFile[]]}*/
       const [processedFiles] = await Promise.all([addVisionPrompt(), getRequestFileIds()]);
@@ -431,7 +400,7 @@ router.post('/', validateModel, buildEndpointOption, setHeaders, async (req, res
 
       if (processedFiles) {
         for (const file of processedFiles) {
-          if (file.source !== FileSources.openai) {
+          if (!checkOpenAIStorage(file.source)) {
             attachedFileIds.delete(file.file_id);
             const index = file_ids.indexOf(file.file_id);
             if (index > -1) {
@@ -467,16 +436,17 @@ router.post('/', validateModel, buildEndpointOption, setHeaders, async (req, res
         assistant_id,
         thread_id,
         model: assistant_id,
+        endpoint,
       };
 
       previousMessages.push(requestMessage);
 
       /* asynchronous */
-      saveUserMessage({ ...requestMessage, model });
+      userMessagePromise = saveUserMessage(req, { ...requestMessage, model });
 
       conversation = {
         conversationId,
-        endpoint: EModelEndpoint.assistants,
+        endpoint,
         promptPrefix: promptPrefix,
         instructions: instructions,
         assistant_id,
@@ -513,7 +483,7 @@ router.post('/', validateModel, buildEndpointOption, setHeaders, async (req, res
     let response;
 
     const processRun = async (retry = false) => {
-      if (req.app.locals[EModelEndpoint.azureOpenAI]?.assistants) {
+      if (endpoint === EModelEndpoint.azureAssistants) {
         body.model = openai._options.model;
         openai.attachedFileIds = attachedFileIds;
         openai.visionPromise = visionPromise;
@@ -603,6 +573,7 @@ router.post('/', validateModel, buildEndpointOption, setHeaders, async (req, res
       assistant_id,
       thread_id,
       model: assistant_id,
+      endpoint,
     };
 
     sendMessage(res, {
@@ -615,7 +586,10 @@ router.post('/', validateModel, buildEndpointOption, setHeaders, async (req, res
     });
     res.end();
 
-    await saveAssistantMessage({ ...responseMessage, model });
+    if (userMessagePromise) {
+      await userMessagePromise;
+    }
+    await saveAssistantMessage(req, { ...responseMessage, model });
 
     if (parentMessageId === Constants.NO_PARENT && !_thread_id) {
       addTitle(req, {
@@ -655,6 +629,6 @@ router.post('/', validateModel, buildEndpointOption, setHeaders, async (req, res
   } catch (error) {
     await handleError(error);
   }
-});
+};
 
-module.exports = router;
+module.exports = chatV1;

api/server/controllers/assistants/chatV2.js

@@ -0,0 +1,500 @@
+const { v4 } = require('uuid');
+const {
+  Time,
+  Constants,
+  RunStatus,
+  CacheKeys,
+  ContentTypes,
+  ToolCallTypes,
+  EModelEndpoint,
+  retrievalMimeTypes,
+  AssistantStreamEvents,
+} = require('librechat-data-provider');
+const {
+  initThread,
+  recordUsage,
+  saveUserMessage,
+  addThreadMetadata,
+  saveAssistantMessage,
+} = require('~/server/services/Threads');
+const { runAssistant, createOnTextProgress } = require('~/server/services/AssistantService');
+const { sendMessage, sleep, isEnabled, countTokens } = require('~/server/utils');
+const { createErrorHandler } = require('~/server/controllers/assistants/errors');
+const validateAuthor = require('~/server/middleware/assistants/validateAuthor');
+const { createRun, StreamRunManager } = require('~/server/services/Runs');
+const { addTitle } = require('~/server/services/Endpoints/assistants');
+const { getTransactions } = require('~/models/Transaction');
+const checkBalance = require('~/models/checkBalance');
+const { getConvo } = require('~/models/Conversation');
+const getLogStores = require('~/cache/getLogStores');
+const { getModelMaxTokens } = require('~/utils');
+const { getOpenAIClient } = require('./helpers');
+const { logger } = require('~/config');
+
+const ten_minutes = 1000 * 60 * 10;
+
+/**
+ * @route POST /
+ * @desc Chat with an assistant
+ * @access Public
+ * @param {Express.Request} req - The request object, containing the request data.
+ * @param {Express.Response} res - The response object, used to send back a response.
+ * @returns {void}
+ */
+const chatV2 = async (req, res) => {
+  logger.debug('[/assistants/chat/] req.body', req.body);
+
+  /** @type {{files: MongoFile[]}} */
+  const {
+    text,
+    model,
+    endpoint,
+    files = [],
+    promptPrefix,
+    assistant_id,
+    instructions,
+    thread_id: _thread_id,
+    messageId: _messageId,
+    conversationId: convoId,
+    parentMessageId: _parentId = Constants.NO_PARENT,
+  } = req.body;
+
+  /** @type {OpenAIClient} */
+  let openai;
+  /** @type {string|undefined} - the current thread id */
+  let thread_id = _thread_id;
+  /** @type {string|undefined} - the current run id */
+  let run_id;
+  /** @type {string|undefined} - the parent messageId */
+  let parentMessageId = _parentId;
+  /** @type {TMessage[]} */
+  let previousMessages = [];
+  /** @type {import('librechat-data-provider').TConversation | null} */
+  let conversation = null;
+  /** @type {string[]} */
+  let file_ids = [];
+  /** @type {Set<string>} */
+  let attachedFileIds = new Set();
+  /** @type {TMessage | null} */
+  let requestMessage = null;
+
+  const userMessageId = v4();
+  const responseMessageId = v4();
+
+  /** @type {string} - The conversation UUID - created if undefined */
+  const conversationId = convoId ?? v4();
+
+  const cache = getLogStores(CacheKeys.ABORT_KEYS);
+  const cacheKey = `${req.user.id}:${conversationId}`;
+
+  /** @type {Run | undefined} - The completed run, undefined if incomplete */
+  let completedRun;
+
+  const getContext = () => ({
+    openai,
+    run_id,
+    endpoint,
+    cacheKey,
+    thread_id,
+    completedRun,
+    assistant_id,
+    conversationId,
+    parentMessageId,
+    responseMessageId,
+  });
+
+  const handleError = createErrorHandler({ req, res, getContext });
+
+  try {
+    res.on('close', async () => {
+      if (!completedRun) {
+        await handleError(new Error('Request closed'));
+      }
+    });
+
+    if (convoId && !_thread_id) {
+      completedRun = true;
+      throw new Error('Missing thread_id for existing conversation');
+    }
+
+    if (!assistant_id) {
+      completedRun = true;
+      throw new Error('Missing assistant_id');
+    }
+
+    const checkBalanceBeforeRun = async () => {
+      if (!isEnabled(process.env.CHECK_BALANCE)) {
+        return;
+      }
+      const transactions =
+        (await getTransactions({
+          user: req.user.id,
+          context: 'message',
+          conversationId,
+        })) ?? [];
+
+      const totalPreviousTokens = Math.abs(
+        transactions.reduce((acc, curr) => acc + curr.rawAmount, 0),
+      );
+
+      // TODO: make promptBuffer a config option; buffer for titles, needs buffer for system instructions
+      const promptBuffer = parentMessageId === Constants.NO_PARENT && !_thread_id ? 200 : 0;
+      // 5 is added for labels
+      let promptTokens = (await countTokens(text + (promptPrefix ?? ''))) + 5;
+      promptTokens += totalPreviousTokens + promptBuffer;
+      // Count tokens up to the current context window
+      promptTokens = Math.min(promptTokens, getModelMaxTokens(model));
+
+      await checkBalance({
+        req,
+        res,
+        txData: {
+          model,
+          user: req.user.id,
+          tokenType: 'prompt',
+          amount: promptTokens,
+        },
+      });
+    };
+
+    const { openai: _openai, client } = await getOpenAIClient({
+      req,
+      res,
+      endpointOption: req.body.endpointOption,
+      initAppClient: true,
+    });
+
+    openai = _openai;
+    await validateAuthor({ req, openai });
+
+    if (previousMessages.length) {
+      parentMessageId = previousMessages[previousMessages.length - 1].messageId;
+    }
+
+    let userMessage = {
+      role: 'user',
+      content: [
+        {
+          type: ContentTypes.TEXT,
+          text,
+        },
+      ],
+      metadata: {
+        messageId: userMessageId,
+      },
+    };
+
+    /** @type {CreateRunBody | undefined} */
+    const body = {
+      assistant_id,
+      model,
+    };
+
+    if (promptPrefix) {
+      body.additional_instructions = promptPrefix;
+    }
+
+    if (instructions) {
+      body.instructions = instructions;
+    }
+
+    const getRequestFileIds = async () => {
+      let thread_file_ids = [];
+      if (convoId) {
+        const convo = await getConvo(req.user.id, convoId);
+        if (convo && convo.file_ids) {
+          thread_file_ids = convo.file_ids;
+        }
+      }
+
+      if (files.length || thread_file_ids.length) {
+        attachedFileIds = new Set([...file_ids, ...thread_file_ids]);
+
+        let attachmentIndex = 0;
+        for (const file of files) {
+          file_ids.push(file.file_id);
+          if (file.type.startsWith('image')) {
+            userMessage.content.push({
+              type: ContentTypes.IMAGE_FILE,
+              [ContentTypes.IMAGE_FILE]: { file_id: file.file_id },
+            });
+          }
+
+          if (!userMessage.attachments) {
+            userMessage.attachments = [];
+          }
+
+          userMessage.attachments.push({
+            file_id: file.file_id,
+            tools: [{ type: ToolCallTypes.CODE_INTERPRETER }],
+          });
+
+          if (file.type.startsWith('image')) {
+            continue;
+          }
+
+          const mimeType = file.type;
+          const isSupportedByRetrieval = retrievalMimeTypes.some((regex) => regex.test(mimeType));
+          if (isSupportedByRetrieval) {
+            userMessage.attachments[attachmentIndex].tools.push({
+              type: ToolCallTypes.FILE_SEARCH,
+            });
+          }
+
+          attachmentIndex++;
+        }
+      }
+    };
+
+    /** @type {Promise<Run>|undefined} */
+    let userMessagePromise;
+
+    const initializeThread = async () => {
+      await getRequestFileIds();
+
+      // TODO: may allow multiple messages to be created beforehand in a future update
+      const initThreadBody = {
+        messages: [userMessage],
+        metadata: {
+          user: req.user.id,
+          conversationId,
+        },
+      };
+
+      const result = await initThread({ openai, body: initThreadBody, thread_id });
+      thread_id = result.thread_id;
+
+      createOnTextProgress({
+        openai,
+        conversationId,
+        userMessageId,
+        messageId: responseMessageId,
+        thread_id,
+      });
+
+      requestMessage = {
+        user: req.user.id,
+        text,
+        messageId: userMessageId,
+        parentMessageId,
+        // TODO: make sure client sends correct format for `files`, use zod
+        files,
+        file_ids,
+        conversationId,
+        isCreatedByUser: true,
+        assistant_id,
+        thread_id,
+        model: assistant_id,
+        endpoint,
+      };
+
+      previousMessages.push(requestMessage);
+
+      /* asynchronous */
+      userMessagePromise = saveUserMessage(req, { ...requestMessage, model });
+
+      conversation = {
+        conversationId,
+        endpoint,
+        promptPrefix: promptPrefix,
+        instructions: instructions,
+        assistant_id,
+        // model,
+      };
+
+      if (file_ids.length) {
+        conversation.file_ids = file_ids;
+      }
+    };
+
+    const promises = [initializeThread(), checkBalanceBeforeRun()];
+    await Promise.all(promises);
+
+    const sendInitialResponse = () => {
+      sendMessage(res, {
+        sync: true,
+        conversationId,
+        // messages: previousMessages,
+        requestMessage,
+        responseMessage: {
+          user: req.user.id,
+          messageId: openai.responseMessage.messageId,
+          parentMessageId: userMessageId,
+          conversationId,
+          assistant_id,
+          thread_id,
+          model: assistant_id,
+        },
+      });
+    };
+
+    /** @type {RunResponse | typeof StreamRunManager | undefined} */
+    let response;
+
+    const processRun = async (retry = false) => {
+      if (endpoint === EModelEndpoint.azureAssistants) {
+        body.model = openai._options.model;
+        openai.attachedFileIds = attachedFileIds;
+        if (retry) {
+          response = await runAssistant({
+            openai,
+            thread_id,
+            run_id,
+            in_progress: openai.in_progress,
+          });
+          return;
+        }
+
+        /* NOTE:
+         * By default, a Run will use the model and tools configuration specified in Assistant object,
+         * but you can override most of these when creating the Run for added flexibility:
+         */
+        const run = await createRun({
+          openai,
+          thread_id,
+          body,
+        });
+
+        run_id = run.id;
+        await cache.set(cacheKey, `${thread_id}:${run_id}`, ten_minutes);
+        sendInitialResponse();
+
+        // todo: retry logic
+        response = await runAssistant({ openai, thread_id, run_id });
+        return;
+      }
+
+      /** @type {{[AssistantStreamEvents.ThreadRunCreated]: (event: ThreadRunCreated) => Promise<void>}} */
+      const handlers = {
+        [AssistantStreamEvents.ThreadRunCreated]: async (event) => {
+          await cache.set(cacheKey, `${thread_id}:${event.data.id}`, ten_minutes);
+          run_id = event.data.id;
+          sendInitialResponse();
+        },
+      };
+
+      /** @type {undefined | TAssistantEndpoint} */
+      const config = req.app.locals[endpoint] ?? {};
+      /** @type {undefined | TBaseEndpoint} */
+      const allConfig = req.app.locals.all;
+
+      const streamRunManager = new StreamRunManager({
+        req,
+        res,
+        openai,
+        handlers,
+        thread_id,
+        attachedFileIds,
+        parentMessageId: userMessageId,
+        responseMessage: openai.responseMessage,
+        streamRate: allConfig?.streamRate ?? config.streamRate,
+        // streamOptions: {
+
+        // },
+      });
+
+      await streamRunManager.runAssistant({
+        thread_id,
+        body,
+      });
+
+      response = streamRunManager;
+      response.text = streamRunManager.intermediateText;
+
+      const messageCache = getLogStores(CacheKeys.MESSAGES);
+      messageCache.set(
+        responseMessageId,
+        {
+          complete: true,
+          text: response.text,
+        },
+        Time.FIVE_MINUTES,
+      );
+    };
+
+    await processRun();
+    logger.debug('[/assistants/chat/] response', {
+      run: response.run,
+      steps: response.steps,
+    });
+
+    if (response.run.status === RunStatus.CANCELLED) {
+      logger.debug('[/assistants/chat/] Run cancelled, handled by `abortRun`');
+      return res.end();
+    }
+
+    if (response.run.status === RunStatus.IN_PROGRESS) {
+      processRun(true);
+    }
+
+    completedRun = response.run;
+
+    /** @type {ResponseMessage} */
+    const responseMessage = {
+      ...(response.responseMessage ?? response.finalMessage),
+      text: response.text,
+      parentMessageId: userMessageId,
+      conversationId,
+      user: req.user.id,
+      assistant_id,
+      thread_id,
+      model: assistant_id,
+      endpoint,
+    };
+
+    sendMessage(res, {
+      final: true,
+      conversation,
+      requestMessage: {
+        parentMessageId,
+        thread_id,
+      },
+    });
+    res.end();
+
+    if (userMessagePromise) {
+      await userMessagePromise;
+    }
+    await saveAssistantMessage(req, { ...responseMessage, model });
+
+    if (parentMessageId === Constants.NO_PARENT && !_thread_id) {
+      addTitle(req, {
+        text,
+        responseText: response.text,
+        conversationId,
+        client,
+      });
+    }
+
+    await addThreadMetadata({
+      openai,
+      thread_id,
+      messageId: responseMessage.messageId,
+      messages: response.messages,
+    });
+
+    if (!response.run.usage) {
+      await sleep(3000);
+      completedRun = await openai.beta.threads.runs.retrieve(thread_id, response.run.id);
+      if (completedRun.usage) {
+        await recordUsage({
+          ...completedRun.usage,
+          user: req.user.id,
+          model: completedRun.model ?? model,
+          conversationId,
+        });
+      }
+    } else {
+      await recordUsage({
+        ...response.run.usage,
+        user: req.user.id,
+        model: response.run.model ?? model,
+        conversationId,
+      });
+    }
+  } catch (error) {
+    await handleError(error);
+  }
+};
+
+module.exports = chatV2;

api/server/controllers/assistants/errors.js

@@ -0,0 +1,193 @@
+// errorHandler.js
+const { sendResponse } = require('~/server/utils');
+const { logger } = require('~/config');
+const getLogStores = require('~/cache/getLogStores');
+const { CacheKeys, ViolationTypes, ContentTypes } = require('librechat-data-provider');
+const { getConvo } = require('~/models/Conversation');
+const { recordUsage, checkMessageGaps } = require('~/server/services/Threads');
+
+/**
+ * @typedef {Object} ErrorHandlerContext
+ * @property {OpenAIClient} openai - The OpenAI client
+ * @property {string} thread_id - The thread ID
+ * @property {string} run_id - The run ID
+ * @property {boolean} completedRun - Whether the run has completed
+ * @property {string} assistant_id - The assistant ID
+ * @property {string} conversationId - The conversation ID
+ * @property {string} parentMessageId - The parent message ID
+ * @property {string} responseMessageId - The response message ID
+ * @property {string} endpoint - The endpoint being used
+ * @property {string} cacheKey - The cache key for the current request
+ */
+
+/**
+ * @typedef {Object} ErrorHandlerDependencies
+ * @property {Express.Request} req - The Express request object
+ * @property {Express.Response} res - The Express response object
+ * @property {() => ErrorHandlerContext} getContext - Function to get the current context
+ * @property {string} [originPath] - The origin path for the error handler
+ */
+
+/**
+ * Creates an error handler function with the given dependencies
+ * @param {ErrorHandlerDependencies} dependencies - The dependencies for the error handler
+ * @returns {(error: Error) => Promise<void>} The error handler function
+ */
+const createErrorHandler = ({ req, res, getContext, originPath = '/assistants/chat/' }) => {
+  const cache = getLogStores(CacheKeys.ABORT_KEYS);
+
+  /**
+   * Handles errors that occur during the chat process
+   * @param {Error} error - The error that occurred
+   * @returns {Promise<void>}
+   */
+  return async (error) => {
+    const {
+      openai,
+      run_id,
+      endpoint,
+      cacheKey,
+      thread_id,
+      completedRun,
+      assistant_id,
+      conversationId,
+      parentMessageId,
+      responseMessageId,
+    } = getContext();
+
+    const defaultErrorMessage =
+      'The Assistant run failed to initialize. Try sending a message in a new conversation.';
+    const messageData = {
+      thread_id,
+      assistant_id,
+      conversationId,
+      parentMessageId,
+      sender: 'System',
+      user: req.user.id,
+      shouldSaveMessage: false,
+      messageId: responseMessageId,
+      endpoint,
+    };
+
+    if (error.message === 'Run cancelled') {
+      return res.end();
+    } else if (error.message === 'Request closed' && completedRun) {
+      return;
+    } else if (error.message === 'Request closed') {
+      logger.debug(`[${originPath}] Request aborted on close`);
+    } else if (/Files.*are invalid/.test(error.message)) {
+      const errorMessage = `Files are invalid, or may not have uploaded yet.${
+        endpoint === 'azureAssistants'
+          ? ' If using Azure OpenAI, files are only available in the region of the assistant\'s model at the time of upload.'
+          : ''
+      }`;
+      return sendResponse(req, res, messageData, errorMessage);
+    } else if (error?.message?.includes('string too long')) {
+      return sendResponse(
+        req,
+        res,
+        messageData,
+        'Message too long. The Assistants API has a limit of 32,768 characters per message. Please shorten it and try again.',
+      );
+    } else if (error?.message?.includes(ViolationTypes.TOKEN_BALANCE)) {
+      return sendResponse(req, res, messageData, error.message);
+    } else {
+      logger.error(`[${originPath}]`, error);
+    }
+
+    if (!openai || !thread_id || !run_id) {
+      return sendResponse(req, res, messageData, defaultErrorMessage);
+    }
+
+    await new Promise((resolve) => setTimeout(resolve, 2000));
+
+    try {
+      const status = await cache.get(cacheKey);
+      if (status === 'cancelled') {
+        logger.debug(`[${originPath}] Run already cancelled`);
+        return res.end();
+      }
+      await cache.delete(cacheKey);
+      const cancelledRun = await openai.beta.threads.runs.cancel(thread_id, run_id);
+      logger.debug(`[${originPath}] Cancelled run:`, cancelledRun);
+    } catch (error) {
+      logger.error(`[${originPath}] Error cancelling run`, error);
+    }
+
+    await new Promise((resolve) => setTimeout(resolve, 2000));
+
+    let run;
+    try {
+      run = await openai.beta.threads.runs.retrieve(thread_id, run_id);
+      await recordUsage({
+        ...run.usage,
+        model: run.model,
+        user: req.user.id,
+        conversationId,
+      });
+    } catch (error) {
+      logger.error(`[${originPath}] Error fetching or processing run`, error);
+    }
+
+    let finalEvent;
+    try {
+      const runMessages = await checkMessageGaps({
+        openai,
+        run_id,
+        endpoint,
+        thread_id,
+        conversationId,
+        latestMessageId: responseMessageId,
+      });
+
+      const errorContentPart = {
+        text: {
+          value:
+            error?.message ?? 'There was an error processing your request. Please try again later.',
+        },
+        type: ContentTypes.ERROR,
+      };
+
+      if (!Array.isArray(runMessages[runMessages.length - 1]?.content)) {
+        runMessages[runMessages.length - 1].content = [errorContentPart];
+      } else {
+        const contentParts = runMessages[runMessages.length - 1].content;
+        for (let i = 0; i < contentParts.length; i++) {
+          const currentPart = contentParts[i];
+          /** @type {CodeToolCall | RetrievalToolCall | FunctionToolCall | undefined} */
+          const toolCall = currentPart?.[ContentTypes.TOOL_CALL];
+          if (
+            toolCall &&
+            toolCall?.function &&
+            !(toolCall?.function?.output || toolCall?.function?.output?.length)
+          ) {
+            contentParts[i] = {
+              ...currentPart,
+              [ContentTypes.TOOL_CALL]: {
+                ...toolCall,
+                function: {
+                  ...toolCall.function,
+                  output: 'error processing tool',
+                },
+              },
+            };
+          }
+        }
+        runMessages[runMessages.length - 1].content.push(errorContentPart);
+      }
+
+      finalEvent = {
+        final: true,
+        conversation: await getConvo(req.user.id, conversationId),
+        runMessages,
+      };
+    } catch (error) {
+      logger.error(`[${originPath}] Error finalizing error process`, error);
+      return sendResponse(req, res, messageData, 'The Assistant run failed');
+    }
+
+    return sendResponse(req, res, finalEvent);
+  };
+};
+
+module.exports = { createErrorHandler };

api/server/controllers/assistants/helpers.js

@@ -0,0 +1,270 @@
+const {
+  CacheKeys,
+  SystemRoles,
+  EModelEndpoint,
+  defaultOrderQuery,
+  defaultAssistantsVersion,
+} = require('librechat-data-provider');
+const {
+  initializeClient: initAzureClient,
+} = require('~/server/services/Endpoints/azureAssistants');
+const { initializeClient } = require('~/server/services/Endpoints/assistants');
+const { getLogStores } = require('~/cache');
+
+/**
+ * @param {Express.Request} req
+ * @param {string} [endpoint]
+ * @returns {Promise<string>}
+ */
+const getCurrentVersion = async (req, endpoint) => {
+  const index = req.baseUrl.lastIndexOf('/v');
+  let version = index !== -1 ? req.baseUrl.substring(index + 1, index + 3) : null;
+  if (!version && req.body.version) {
+    version = `v${req.body.version}`;
+  }
+  if (!version && endpoint) {
+    const cache = getLogStores(CacheKeys.CONFIG_STORE);
+    const cachedEndpointsConfig = await cache.get(CacheKeys.ENDPOINT_CONFIG);
+    version = `v${
+      cachedEndpointsConfig?.[endpoint]?.version ?? defaultAssistantsVersion[endpoint]
+    }`;
+  }
+  if (!version?.startsWith('v') && version.length !== 2) {
+    throw new Error(`[${req.baseUrl}] Invalid version: ${version}`);
+  }
+  return version;
+};
+
+/**
+ * Asynchronously lists assistants based on provided query parameters.
+ *
+ * Initializes the client with the current request and response objects and lists assistants
+ * according to the query parameters. This function abstracts the logic for non-Azure paths.
+ *
+ * @deprecated
+ * @async
+ * @param {object} params - The parameters object.
+ * @param {object} params.req - The request object, used for initializing the client.
+ * @param {object} params.res - The response object, used for initializing the client.
+ * @param {string} params.version - The API version to use.
+ * @param {object} params.query - The query parameters to list assistants (e.g., limit, order).
+ * @returns {Promise<object>} A promise that resolves to the response from the `openai.beta.assistants.list` method call.
+ */
+const _listAssistants = async ({ req, res, version, query }) => {
+  const { openai } = await getOpenAIClient({ req, res, version });
+  return openai.beta.assistants.list(query);
+};
+
+/**
+ * Fetches all assistants based on provided query params, until `has_more` is `false`.
+ *
+ * @async
+ * @param {object} params - The parameters object.
+ * @param {object} params.req - The request object, used for initializing the client.
+ * @param {object} params.res - The response object, used for initializing the client.
+ * @param {string} params.version - The API version to use.
+ * @param {Omit<AssistantListParams, 'endpoint'>} params.query - The query parameters to list assistants (e.g., limit, order).
+ * @returns {Promise<object>} A promise that resolves to the response from the `openai.beta.assistants.list` method call.
+ */
+const listAllAssistants = async ({ req, res, version, query }) => {
+  /** @type {{ openai: OpenAIClient }} */
+  const { openai } = await getOpenAIClient({ req, res, version });
+  const allAssistants = [];
+
+  let first_id;
+  let last_id;
+  let afterToken = query.after;
+  let hasMore = true;
+
+  while (hasMore) {
+    const response = await openai.beta.assistants.list({
+      ...query,
+      after: afterToken,
+    });
+
+    const { body } = response;
+
+    allAssistants.push(...body.data);
+    hasMore = body.has_more;
+
+    if (!first_id) {
+      first_id = body.first_id;
+    }
+
+    if (hasMore) {
+      afterToken = body.last_id;
+    } else {
+      last_id = body.last_id;
+    }
+  }
+
+  return {
+    data: allAssistants,
+    body: {
+      data: allAssistants,
+      has_more: false,
+      first_id,
+      last_id,
+    },
+  };
+};
+
+/**
+ * Asynchronously lists assistants for Azure configured groups.
+ *
+ * Iterates through Azure configured assistant groups, initializes the client with the current request and response objects,
+ * lists assistants based on the provided query parameters, and merges their data alongside the model information into a single array.
+ *
+ * @async
+ * @param {object} params - The parameters object.
+ * @param {object} params.req - The request object, used for initializing the client and manipulating the request body.
+ * @param {object} params.res - The response object, used for initializing the client.
+ * @param {string} params.version - The API version to use.
+ * @param {TAzureConfig} params.azureConfig - The Azure configuration object containing assistantGroups and groupMap.
+ * @param {object} params.query - The query parameters to list assistants (e.g., limit, order).
+ * @returns {Promise<AssistantListResponse>} A promise that resolves to an array of assistant data merged with their respective model information.
+ */
+const listAssistantsForAzure = async ({ req, res, version, azureConfig = {}, query }) => {
+  /** @type {Array<[string, TAzureModelConfig]>} */
+  const groupModelTuples = [];
+  const promises = [];
+  /** @type {Array<TAzureGroup>} */
+  const groups = [];
+
+  const { groupMap, assistantGroups } = azureConfig;
+
+  for (const groupName of assistantGroups) {
+    const group = groupMap[groupName];
+    groups.push(group);
+
+    const currentModelTuples = Object.entries(group?.models);
+    groupModelTuples.push(currentModelTuples);
+
+    /* The specified model is only necessary to
+    fetch assistants for the shared instance */
+    req.body.model = currentModelTuples[0][0];
+    promises.push(listAllAssistants({ req, res, version, query }));
+  }
+
+  const resolvedQueries = await Promise.all(promises);
+  const data = resolvedQueries.flatMap((res, i) =>
+    res.data.map((assistant) => {
+      const deploymentName = assistant.model;
+      const currentGroup = groups[i];
+      const currentModelTuples = groupModelTuples[i];
+      const firstModel = currentModelTuples[0][0];
+
+      if (currentGroup.deploymentName === deploymentName) {
+        return { ...assistant, model: firstModel };
+      }
+
+      for (const [model, modelConfig] of currentModelTuples) {
+        if (modelConfig.deploymentName === deploymentName) {
+          return { ...assistant, model };
+        }
+      }
+
+      return { ...assistant, model: firstModel };
+    }),
+  );
+
+  return {
+    first_id: data[0]?.id,
+    last_id: data[data.length - 1]?.id,
+    object: 'list',
+    has_more: false,
+    data,
+  };
+};
+
+async function getOpenAIClient({ req, res, endpointOption, initAppClient, overrideEndpoint }) {
+  let endpoint = overrideEndpoint ?? req.body.endpoint ?? req.query.endpoint;
+  const version = await getCurrentVersion(req, endpoint);
+  if (!endpoint) {
+    throw new Error(`[${req.baseUrl}] Endpoint is required`);
+  }
+
+  let result;
+  if (endpoint === EModelEndpoint.assistants) {
+    result = await initializeClient({ req, res, version, endpointOption, initAppClient });
+  } else if (endpoint === EModelEndpoint.azureAssistants) {
+    result = await initAzureClient({ req, res, version, endpointOption, initAppClient });
+  }
+
+  return result;
+}
+
+/**
+ * Returns a list of assistants.
+ * @param {object} params
+ * @param {object} params.req - Express Request
+ * @param {AssistantListParams} [params.req.query] - The assistant list parameters for pagination and sorting.
+ * @param {object} params.res - Express Response
+ * @param {string} [params.overrideEndpoint] - The endpoint to override the request endpoint.
+ * @returns {Promise<AssistantListResponse>} 200 - success response - application/json
+ */
+const fetchAssistants = async ({ req, res, overrideEndpoint }) => {
+  const {
+    limit = 100,
+    order = 'desc',
+    after,
+    before,
+    endpoint,
+  } = req.query ?? {
+    endpoint: overrideEndpoint,
+    ...defaultOrderQuery,
+  };
+
+  const version = await getCurrentVersion(req, endpoint);
+  const query = { limit, order, after, before };
+
+  /** @type {AssistantListResponse} */
+  let body;
+
+  if (endpoint === EModelEndpoint.assistants) {
+    ({ body } = await listAllAssistants({ req, res, version, query }));
+  } else if (endpoint === EModelEndpoint.azureAssistants) {
+    const azureConfig = req.app.locals[EModelEndpoint.azureOpenAI];
+    body = await listAssistantsForAzure({ req, res, version, azureConfig, query });
+  }
+
+  if (req.user.role === SystemRoles.ADMIN) {
+    return body;
+  } else if (!req.app.locals[endpoint]) {
+    return body;
+  }
+
+  body.data = filterAssistants({
+    userId: req.user.id,
+    assistants: body.data,
+    assistantsConfig: req.app.locals[endpoint],
+  });
+  return body;
+};
+
+/**
+ * Filter assistants based on configuration.
+ *
+ * @param {object} params - The parameters object.
+ * @param {string} params.userId -  The user ID to filter private assistants.
+ * @param {Assistant[]} params.assistants - The list of assistants to filter.
+ * @param {Partial<TAssistantEndpoint>} params.assistantsConfig -  The assistant configuration.
+ * @returns {Assistant[]} - The filtered list of assistants.
+ */
+function filterAssistants({ assistants, userId, assistantsConfig }) {
+  const { supportedIds, excludedIds, privateAssistants } = assistantsConfig;
+  if (privateAssistants) {
+    return assistants.filter((assistant) => userId === assistant.metadata?.author);
+  } else if (supportedIds?.length) {
+    return assistants.filter((assistant) => supportedIds.includes(assistant.id));
+  } else if (excludedIds?.length) {
+    return assistants.filter((assistant) => !excludedIds.includes(assistant.id));
+  }
+  return assistants;
+}
+
+module.exports = {
+  getOpenAIClient,
+  fetchAssistants,
+  getCurrentVersion,
+};

api/server/controllers/assistants/v1.js

@@ -1,34 +1,12 @@
-const multer = require('multer');
-const express = require('express');
-const { FileContext, EModelEndpoint } = require('librechat-data-provider');
-const {
-  initializeClient,
-  listAssistantsForAzure,
-  listAssistants,
-} = require('~/server/services/Endpoints/assistants');
+const { FileContext } = require('librechat-data-provider');
+const validateAuthor = require('~/server/middleware/assistants/validateAuthor');
 const { getStrategyFunctions } = require('~/server/services/Files/strategies');
+const { deleteAssistantActions } = require('~/server/services/ActionService');
+const { updateAssistantDoc, getAssistants } = require('~/models/Assistant');
 const { uploadImageBuffer } = require('~/server/services/Files/process');
-const { updateAssistant, getAssistants } = require('~/models/Assistant');
+const { getOpenAIClient, fetchAssistants } = require('./helpers');
 const { deleteFileByFilter } = require('~/models/File');
 const { logger } = require('~/config');
-const actions = require('./actions');
-const tools = require('./tools');
-
-const upload = multer();
-const router = express.Router();
-
-/**
- * Assistant actions route.
- * @route GET|POST /assistants/actions
- */
-router.use('/actions', actions);
-
-/**
- * Create an assistant.
- * @route GET /assistants/tools
- * @returns {TPlugin[]} 200 - application/json
- */
-router.use('/tools', tools);
 
 /**
  * Create an assistant.
@@ -36,12 +14,11 @@ router.use('/tools', tools);
  * @param {AssistantCreateParams} req.body - The assistant creation parameters.
  * @returns {Assistant} 201 - success response - application/json
  */
-router.post('/', async (req, res) => {
+const createAssistant = async (req, res) => {
   try {
-    /** @type {{ openai: OpenAI }} */
-    const { openai } = await initializeClient({ req, res });
+    const { openai } = await getOpenAIClient({ req, res });
 
-    const { tools = [], ...assistantData } = req.body;
+    const { tools = [], endpoint, ...assistantData } = req.body;
     assistantData.tools = tools
       .map((tool) => {
         if (typeof tool !== 'string') {
@@ -52,18 +29,30 @@ router.post('/', async (req, res) => {
       })
       .filter((tool) => tool);
 
+    let azureModelIdentifier = null;
     if (openai.locals?.azureOptions) {
+      azureModelIdentifier = assistantData.model;
       assistantData.model = openai.locals.azureOptions.azureOpenAIApiDeploymentName;
     }
 
+    assistantData.metadata = {
+      author: req.user.id,
+      endpoint,
+    };
+
     const assistant = await openai.beta.assistants.create(assistantData);
+    const promise = updateAssistantDoc({ assistant_id: assistant.id }, { user: req.user.id });
+    if (azureModelIdentifier) {
+      assistant.model = azureModelIdentifier;
+    }
+    await promise;
     logger.debug('/assistants/', assistant);
     res.status(201).json(assistant);
   } catch (error) {
     logger.error('[/assistants] Error creating assistant', error);
     res.status(500).json({ error: error.message });
   }
-});
+};
 
 /**
  * Retrieves an assistant.
@@ -71,11 +60,10 @@ router.post('/', async (req, res) => {
  * @param {string} req.params.id - Assistant identifier.
  * @returns {Assistant} 200 - success response - application/json
  */
-router.get('/:id', async (req, res) => {
+const retrieveAssistant = async (req, res) => {
   try {
-    /** @type {{ openai: OpenAI }} */
-    const { openai } = await initializeClient({ req, res });
-
+    /* NOTE: not actually being used right now */
+    const { openai } = await getOpenAIClient({ req, res });
     const assistant_id = req.params.id;
     const assistant = await openai.beta.assistants.retrieve(assistant_id);
     res.json(assistant);
@@ -83,22 +71,24 @@ router.get('/:id', async (req, res) => {
     logger.error('[/assistants/:id] Error retrieving assistant', error);
     res.status(500).json({ error: error.message });
   }
-});
+};
 
 /**
  * Modifies an assistant.
  * @route PATCH /assistants/:id
+ * @param {object} req - Express Request
+ * @param {object} req.params - Request params
  * @param {string} req.params.id - Assistant identifier.
  * @param {AssistantUpdateParams} req.body - The assistant update parameters.
  * @returns {Assistant} 200 - success response - application/json
  */
-router.patch('/:id', async (req, res) => {
+const patchAssistant = async (req, res) => {
   try {
-    /** @type {{ openai: OpenAI }} */
-    const { openai } = await initializeClient({ req, res });
+    const { openai } = await getOpenAIClient({ req, res });
+    await validateAuthor({ req, openai });
 
     const assistant_id = req.params.id;
-    const updateData = req.body;
+    const { endpoint: _e, ...updateData } = req.body;
     updateData.tools = (updateData.tools ?? [])
       .map((tool) => {
         if (typeof tool !== 'string') {
@@ -119,90 +109,76 @@ router.patch('/:id', async (req, res) => {
     logger.error('[/assistants/:id] Error updating assistant', error);
     res.status(500).json({ error: error.message });
   }
-});
+};
 
 /**
  * Deletes an assistant.
  * @route DELETE /assistants/:id
+ * @param {object} req - Express Request
+ * @param {object} req.params - Request params
  * @param {string} req.params.id - Assistant identifier.
  * @returns {Assistant} 200 - success response - application/json
  */
-router.delete('/:id', async (req, res) => {
+const deleteAssistant = async (req, res) => {
   try {
-    /** @type {{ openai: OpenAI }} */
-    const { openai } = await initializeClient({ req, res });
+    const { openai } = await getOpenAIClient({ req, res });
+    await validateAuthor({ req, openai });
 
     const assistant_id = req.params.id;
     const deletionStatus = await openai.beta.assistants.del(assistant_id);
+    if (deletionStatus?.deleted) {
+      await deleteAssistantActions({ req, assistant_id });
+    }
     res.json(deletionStatus);
   } catch (error) {
     logger.error('[/assistants/:id] Error deleting assistant', error);
     res.status(500).json({ error: 'Error deleting assistant' });
   }
-});
+};
 
 /**
  * Returns a list of assistants.
  * @route GET /assistants
+ * @param {object} req - Express Request
  * @param {AssistantListParams} req.query - The assistant list parameters for pagination and sorting.
  * @returns {AssistantListResponse} 200 - success response - application/json
  */
-router.get('/', async (req, res) => {
+const listAssistants = async (req, res) => {
   try {
-    const { limit = 100, order = 'desc', after, before } = req.query;
-    const query = { limit, order, after, before };
-
-    const azureConfig = req.app.locals[EModelEndpoint.azureOpenAI];
-    /** @type {AssistantListResponse} */
-    let body;
-
-    if (azureConfig?.assistants) {
-      body = await listAssistantsForAzure({ req, res, azureConfig, query });
-    } else {
-      ({ body } = await listAssistants({ req, res, query }));
-    }
-
-    if (req.app.locals?.[EModelEndpoint.assistants]) {
-      /** @type {Partial<TAssistantEndpoint>} */
-      const assistantsConfig = req.app.locals[EModelEndpoint.assistants];
-      const { supportedIds, excludedIds } = assistantsConfig;
-      if (supportedIds?.length) {
-        body.data = body.data.filter((assistant) => supportedIds.includes(assistant.id));
-      } else if (excludedIds?.length) {
-        body.data = body.data.filter((assistant) => !excludedIds.includes(assistant.id));
-      }
-    }
-
+    const body = await fetchAssistants({ req, res });
     res.json(body);
   } catch (error) {
     logger.error('[/assistants] Error listing assistants', error);
     res.status(500).json({ message: 'Error listing assistants' });
   }
-});
+};
 
 /**
  * Returns a list of the user's assistant documents (metadata saved to database).
  * @route GET /assistants/documents
  * @returns {AssistantDocument[]} 200 - success response - application/json
  */
-router.get('/documents', async (req, res) => {
+const getAssistantDocuments = async (req, res) => {
   try {
     res.json(await getAssistants({ user: req.user.id }));
   } catch (error) {
     logger.error('[/assistants/documents] Error listing assistant documents', error);
     res.status(500).json({ error: error.message });
   }
-});
+};
 
 /**
  * Uploads and updates an avatar for a specific assistant.
  * @route POST /avatar/:assistant_id
+ * @param {object} req - Express Request
+ * @param {object} req.params - Request params
  * @param {string} req.params.assistant_id - The ID of the assistant.
  * @param {Express.Multer.File} req.file - The avatar image file.
+ * @param {object} req.body - Request body
  * @param {string} [req.body.metadata] - Optional metadata for the assistant's avatar.
  * @returns {Object} 200 - success response - application/json
  */
-router.post('/avatar/:assistant_id', upload.single('file'), async (req, res) => {
+const uploadAssistantAvatar = async (req, res) => {
   try {
     const { assistant_id } = req.params;
     if (!assistant_id) {
@@ -210,8 +186,8 @@ router.post('/avatar/:assistant_id', upload.single('file'), async (req, res) =>
     }
 
     let { metadata: _metadata = '{}' } = req.body;
-    /** @type {{ openai: OpenAI }} */
-    const { openai } = await initializeClient({ req, res });
+    const { openai } = await getOpenAIClient({ req, res });
+    await validateAuthor({ req, openai });
 
     const image = await uploadImageBuffer({
       req,
@@ -246,7 +222,7 @@ router.post('/avatar/:assistant_id', upload.single('file'), async (req, res) =>
 
     const promises = [];
     promises.push(
-      updateAssistant(
+      updateAssistantDoc(
         { assistant_id },
         {
           avatar: {
@@ -266,6 +242,14 @@ router.post('/avatar/:assistant_id', upload.single('file'), async (req, res) =>
     logger.error(message, error);
     res.status(500).json({ message });
   }
-});
+};
 
-module.exports = router;
+module.exports = {
+  createAssistant,
+  retrieveAssistant,
+  patchAssistant,
+  deleteAssistant,
+  listAssistants,
+  getAssistantDocuments,
+  uploadAssistantAvatar,
+};

api/server/controllers/assistants/v2.js

@@ -0,0 +1,213 @@
+const { ToolCallTypes } = require('librechat-data-provider');
+const validateAuthor = require('~/server/middleware/assistants/validateAuthor');
+const { validateAndUpdateTool } = require('~/server/services/ActionService');
+const { updateAssistantDoc } = require('~/models/Assistant');
+const { getOpenAIClient } = require('./helpers');
+const { logger } = require('~/config');
+
+/**
+ * Create an assistant.
+ * @route POST /assistants
+ * @param {AssistantCreateParams} req.body - The assistant creation parameters.
+ * @returns {Assistant} 201 - success response - application/json
+ */
+const createAssistant = async (req, res) => {
+  try {
+    /** @type {{ openai: OpenAIClient }} */
+    const { openai } = await getOpenAIClient({ req, res });
+
+    const { tools = [], endpoint, ...assistantData } = req.body;
+    assistantData.tools = tools
+      .map((tool) => {
+        if (typeof tool !== 'string') {
+          return tool;
+        }
+
+        return req.app.locals.availableTools[tool];
+      })
+      .filter((tool) => tool);
+
+    let azureModelIdentifier = null;
+    if (openai.locals?.azureOptions) {
+      azureModelIdentifier = assistantData.model;
+      assistantData.model = openai.locals.azureOptions.azureOpenAIApiDeploymentName;
+    }
+
+    assistantData.metadata = {
+      author: req.user.id,
+      endpoint,
+    };
+
+    const assistant = await openai.beta.assistants.create(assistantData);
+    const promise = updateAssistantDoc({ assistant_id: assistant.id }, { user: req.user.id });
+    if (azureModelIdentifier) {
+      assistant.model = azureModelIdentifier;
+    }
+    await promise;
+    logger.debug('/assistants/', assistant);
+    res.status(201).json(assistant);
+  } catch (error) {
+    logger.error('[/assistants] Error creating assistant', error);
+    res.status(500).json({ error: error.message });
+  }
+};
+
+/**
+ * Modifies an assistant.
+ * @param {object} params
+ * @param {Express.Request} params.req
+ * @param {OpenAIClient} params.openai
+ * @param {string} params.assistant_id
+ * @param {AssistantUpdateParams} params.updateData
+ * @returns {Promise<Assistant>} The updated assistant.
+ */
+const updateAssistant = async ({ req, openai, assistant_id, updateData }) => {
+  await validateAuthor({ req, openai });
+  const tools = [];
+
+  let hasFileSearch = false;
+  for (const tool of updateData.tools ?? []) {
+    let actualTool = typeof tool === 'string' ? req.app.locals.availableTools[tool] : tool;
+
+    if (!actualTool) {
+      continue;
+    }
+
+    if (actualTool.type === ToolCallTypes.FILE_SEARCH) {
+      hasFileSearch = true;
+    }
+
+    if (!actualTool.function) {
+      tools.push(actualTool);
+      continue;
+    }
+
+    const updatedTool = await validateAndUpdateTool({ req, tool: actualTool, assistant_id });
+    if (updatedTool) {
+      tools.push(updatedTool);
+    }
+  }
+
+  if (hasFileSearch && !updateData.tool_resources) {
+    const assistant = await openai.beta.assistants.retrieve(assistant_id);
+    updateData.tool_resources = assistant.tool_resources ?? null;
+  }
+
+  if (hasFileSearch && !updateData.tool_resources?.file_search) {
+    updateData.tool_resources = {
+      ...(updateData.tool_resources ?? {}),
+      file_search: {
+        vector_store_ids: [],
+      },
+    };
+  }
+
+  updateData.tools = tools;
+
+  if (openai.locals?.azureOptions && updateData.model) {
+    updateData.model = openai.locals.azureOptions.azureOpenAIApiDeploymentName;
+  }
+
+  return await openai.beta.assistants.update(assistant_id, updateData);
+};
+
+/**
+ * Modifies an assistant with the resource file id.
+ * @param {object} params
+ * @param {Express.Request} params.req
+ * @param {OpenAIClient} params.openai
+ * @param {string} params.assistant_id
+ * @param {string} params.tool_resource
+ * @param {string} params.file_id
+ * @param {AssistantUpdateParams} params.updateData
+ * @returns {Promise<Assistant>} The updated assistant.
+ */
+const addResourceFileId = async ({ req, openai, assistant_id, tool_resource, file_id }) => {
+  const assistant = await openai.beta.assistants.retrieve(assistant_id);
+  const { tool_resources = {} } = assistant;
+  if (tool_resources[tool_resource]) {
+    tool_resources[tool_resource].file_ids.push(file_id);
+  } else {
+    tool_resources[tool_resource] = { file_ids: [file_id] };
+  }
+
+  delete assistant.id;
+  return await updateAssistant({
+    req,
+    openai,
+    assistant_id,
+    updateData: { tools: assistant.tools, tool_resources },
+  });
+};
+
+/**
+ * Deletes a file ID from an assistant's resource.
+ * @param {object} params
+ * @param {Express.Request} params.req
+ * @param {OpenAIClient} params.openai
+ * @param {string} params.assistant_id
+ * @param {string} [params.tool_resource]
+ * @param {string} params.file_id
+ * @param {AssistantUpdateParams} params.updateData
+ * @returns {Promise<Assistant>} The updated assistant.
+ */
+const deleteResourceFileId = async ({ req, openai, assistant_id, tool_resource, file_id }) => {
+  const assistant = await openai.beta.assistants.retrieve(assistant_id);
+  const { tool_resources = {} } = assistant;
+
+  if (tool_resource && tool_resources[tool_resource]) {
+    const resource = tool_resources[tool_resource];
+    const index = resource.file_ids.indexOf(file_id);
+    if (index !== -1) {
+      resource.file_ids.splice(index, 1);
+    }
+  } else {
+    for (const resourceKey in tool_resources) {
+      const resource = tool_resources[resourceKey];
+      const index = resource.file_ids.indexOf(file_id);
+      if (index !== -1) {
+        resource.file_ids.splice(index, 1);
+        break;
+      }
+    }
+  }
+
+  delete assistant.id;
+  return await updateAssistant({
+    req,
+    openai,
+    assistant_id,
+    updateData: { tools: assistant.tools, tool_resources },
+  });
+};
+
+/**
+ * Modifies an assistant.
+ * @route PATCH /assistants/:id
+ * @param {object} req - Express Request
+ * @param {object} req.params - Request params
+ * @param {string} req.params.id - Assistant identifier.
+ * @param {AssistantUpdateParams} req.body - The assistant update parameters.
+ * @returns {Assistant} 200 - success response - application/json
+ */
+const patchAssistant = async (req, res) => {
+  try {
+    const { openai } = await getOpenAIClient({ req, res });
+    const assistant_id = req.params.id;
+    const { endpoint: _e, ...updateData } = req.body;
+    updateData.tools = updateData.tools ?? [];
+    const updatedAssistant = await updateAssistant({ req, openai, assistant_id, updateData });
+    res.json(updatedAssistant);
+  } catch (error) {
+    logger.error('[/assistants/:id] Error updating assistant', error);
+    res.status(500).json({ error: error.message });
+  }
+};
+
+module.exports = {
+  patchAssistant,
+  createAssistant,
+  updateAssistant,
+  addResourceFileId,
+  deleteResourceFileId,
+};

api/server/controllers/auth/LoginController.js

@@ -1,26 +1,22 @@
-const User = require('~/models/User');
 const { setAuthTokens } = require('~/server/services/AuthService');
 const { logger } = require('~/config');
 
 const loginController = async (req, res) => {
   try {
-    const user = await User.findById(req.user._id);
-
-    // If user doesn't exist, return error
-    if (!user) {
-      // typeof user !== User) { // this doesn't seem to resolve the User type ??
+    if (!req.user) {
       return res.status(400).json({ message: 'Invalid credentials' });
     }
 
-    const token = await setAuthTokens(user._id, res);
+    const { password: _, __v, ...user } = req.user;
+    user.id = user._id.toString();
+
+    const token = await setAuthTokens(req.user._id, res);
 
     return res.status(200).send({ token, user });
   } catch (err) {
     logger.error('[loginController]', err);
+    return res.status(500).json({ message: 'Something went wrong' });
   }
-
-  // Generic error messages are safer
-  return res.status(500).json({ message: 'Something went wrong' });
 };
 
 module.exports = {

api/server/index.js

@@ -4,21 +4,23 @@ require('module-alias')({ base: path.resolve(__dirname, '..') });
 const cors = require('cors');
 const axios = require('axios');
 const express = require('express');
+const compression = require('compression');
 const passport = require('passport');
 const mongoSanitize = require('express-mongo-sanitize');
+const { jwtLogin, passportLogin } = require('~/strategies');
+const { connectDb, indexSync } = require('~/lib/db');
+const { isEnabled } = require('~/server/utils');
+const { ldapLogin } = require('~/strategies');
+const { logger } = require('~/config');
 const validateImageRequest = require('./middleware/validateImageRequest');
 const errorController = require('./controllers/ErrorController');
-const { jwtLogin, passportLogin } = require('~/strategies');
 const configureSocialLogins = require('./socialLogins');
-const { connectDb, indexSync } = require('~/lib/db');
 const AppService = require('./services/AppService');
+const staticCache = require('./utils/staticCache');
 const noIndex = require('./middleware/noIndex');
-const { isEnabled } = require('~/server/utils');
-const { logger } = require('~/config');
-
 const routes = require('./routes');
 
-const { PORT, HOST, ALLOW_SOCIAL_LOGIN } = process.env ?? {};
+const { PORT, HOST, ALLOW_SOCIAL_LOGIN, DISABLE_COMPRESSION } = process.env ?? {};
 
 const port = Number(PORT) || 3080;
 const host = HOST || 'localhost';
@@ -37,35 +39,44 @@ const startServer = async () => {
 
   app.get('/health', (_req, res) => res.status(200).send('OK'));
 
-  // Middleware
+  /* Middleware */
   app.use(noIndex);
   app.use(errorController);
   app.use(express.json({ limit: '3mb' }));
   app.use(mongoSanitize());
   app.use(express.urlencoded({ extended: true, limit: '3mb' }));
-  app.use(express.static(app.locals.paths.dist));
-  app.use(express.static(app.locals.paths.fonts));
-  app.use(express.static(app.locals.paths.assets));
-  app.set('trust proxy', 1); // trust first proxy
+  app.use(staticCache(app.locals.paths.dist));
+  app.use(staticCache(app.locals.paths.fonts));
+  app.use(staticCache(app.locals.paths.assets));
+  app.set('trust proxy', 1); /* trust first proxy */
   app.use(cors());
 
+  if (!isEnabled(DISABLE_COMPRESSION)) {
+    app.use(compression());
+  }
+
   if (!ALLOW_SOCIAL_LOGIN) {
     console.warn(
       'Social logins are disabled. Set Environment Variable "ALLOW_SOCIAL_LOGIN" to true to enable them.',
     );
   }
 
-  // OAUTH
+  /* OAUTH */
   app.use(passport.initialize());
   passport.use(await jwtLogin());
   passport.use(passportLogin());
 
+  /* LDAP Auth */
+  if (process.env.LDAP_URL && process.env.LDAP_USER_SEARCH_BASE) {
+    passport.use(ldapLogin);
+  }
+
   if (isEnabled(ALLOW_SOCIAL_LOGIN)) {
     configureSocialLogins(app);
   }
 
   app.use('/oauth', routes.oauth);
-  // API Endpoints
+  /* API Endpoints */
   app.use('/api/auth', routes.auth);
   app.use('/api/keys', routes.keys);
   app.use('/api/user', routes.user);
@@ -76,6 +87,7 @@ const startServer = async () => {
   app.use('/api/convos', routes.convos);
   app.use('/api/presets', routes.presets);
   app.use('/api/prompts', routes.prompts);
+  app.use('/api/categories', routes.categories);
   app.use('/api/tokenizer', routes.tokenizer);
   app.use('/api/endpoints', routes.endpoints);
   app.use('/api/balance', routes.balance);
@@ -86,9 +98,11 @@ const startServer = async () => {
   app.use('/api/files', await routes.files.initialize());
   app.use('/images/', validateImageRequest, routes.staticRoute);
   app.use('/api/share', routes.share);
+  app.use('/api/roles', routes.roles);
 
+  app.use('/api/tags', routes.tags);
   app.use((req, res) => {
-    res.status(404).sendFile(path.join(app.locals.paths.dist, 'index.html'));
+    res.sendFile(path.join(app.locals.paths.dist, 'index.html'));
   });
 
   app.listen(port, host, () => {

api/server/middleware/abortMiddleware.js

@@ -1,31 +1,39 @@
-const { EModelEndpoint } = require('librechat-data-provider');
+const { isAssistantsEndpoint } = require('librechat-data-provider');
 const { sendMessage, sendError, countTokens, isEnabled } = require('~/server/utils');
 const { truncateText, smartTruncateText } = require('~/app/clients/prompts');
-const { saveMessage, getConvo, getConvoTitle } = require('~/models');
 const clearPendingReq = require('~/cache/clearPendingReq');
+const { spendTokens } = require('~/models/spendTokens');
 const abortControllers = require('./abortControllers');
-const spendTokens = require('~/models/spendTokens');
+const { saveMessage, getConvo } = require('~/models');
 const { abortRun } = require('./abortRun');
 const { logger } = require('~/config');
 
 async function abortMessage(req, res) {
-  let { abortKey, conversationId, endpoint } = req.body;
+  let { abortKey, endpoint } = req.body;
 
-  if (!abortKey && conversationId) {
-    abortKey = conversationId;
+  if (isAssistantsEndpoint(endpoint)) {
+    return await abortRun(req, res);
   }
 
-  if (endpoint === EModelEndpoint.assistants) {
-    return await abortRun(req, res);
+  const conversationId = abortKey?.split(':')?.[0] ?? req.user.id;
+
+  if (!abortControllers.has(abortKey) && abortControllers.has(conversationId)) {
+    abortKey = conversationId;
   }
 
   if (!abortControllers.has(abortKey) && !res.headersSent) {
     return res.status(204).send({ message: 'Request not found' });
   }
 
-  const { abortController } = abortControllers.get(abortKey);
+  const { abortController } = abortControllers.get(abortKey) ?? {};
+  if (!abortController) {
+    return res.status(204).send({ message: 'Request not found' });
+  }
   const finalEvent = await abortController.abortCompletion();
-  logger.debug('[abortMessage] Aborted request', { abortKey });
+  logger.debug(
+    `[abortMessage] ID: ${req.user.id} | ${req.user.email} | Aborted request: ` +
+      JSON.stringify({ abortKey }),
+  );
   abortControllers.delete(abortKey);
 
   if (res.headersSent && finalEvent) {
@@ -50,12 +58,35 @@ const handleAbort = () => {
   };
 };
 
-const createAbortController = (req, res, getAbortData) => {
+const createAbortController = (req, res, getAbortData, getReqData) => {
   const abortController = new AbortController();
   const { endpointOption } = req.body;
-  const onStart = (userMessage) => {
+
+  abortController.getAbortData = function () {
+    return getAbortData();
+  };
+
+  /**
+   * @param {TMessage} userMessage
+   * @param {string} responseMessageId
+   */
+  const onStart = (userMessage, responseMessageId) => {
     sendMessage(res, { message: userMessage, created: true });
+
     const abortKey = userMessage?.conversationId ?? req.user.id;
+    const prevRequest = abortControllers.get(abortKey);
+
+    if (prevRequest && prevRequest?.abortController) {
+      const data = prevRequest.abortController.getAbortData();
+      getReqData({ userMessage: data?.userMessage });
+      const addedAbortKey = `${abortKey}:${responseMessageId}`;
+      abortControllers.set(addedAbortKey, { abortController, ...endpointOption });
+      res.on('finish', function () {
+        abortControllers.delete(addedAbortKey);
+      });
+      return;
+    }
+
     abortControllers.set(abortKey, { abortController, ...endpointOption });
 
     res.on('finish', function () {
@@ -65,7 +96,8 @@ const createAbortController = (req, res, getAbortData) => {
 
   abortController.abortCompletion = async function () {
     abortController.abort();
-    const { conversationId, userMessage, promptTokens, ...responseData } = getAbortData();
+    const { conversationId, userMessage, userMessagePromise, promptTokens, ...responseData } =
+      getAbortData();
     const completionTokens = await countTokens(responseData?.text ?? '');
     const user = req.user.id;
 
@@ -87,12 +119,26 @@ const createAbortController = (req, res, getAbortData) => {
       { promptTokens, completionTokens },
     );
 
-    saveMessage({ ...responseMessage, user });
+    saveMessage(
+      req,
+      { ...responseMessage, user },
+      { context: 'api/server/middleware/abortMiddleware.js' },
+    );
+
+    let conversation;
+    if (userMessagePromise) {
+      const resolved = await userMessagePromise;
+      conversation = resolved?.conversation;
+    }
+
+    if (!conversation) {
+      conversation = await getConvo(req.user.id, conversationId);
+    }
 
     return {
-      title: await getConvoTitle(user, conversationId),
+      title: conversation && !conversation.title ? null : conversation?.title || 'New Chat',
       final: true,
-      conversation: await getConvo(user, conversationId),
+      conversation,
       requestMessage: userMessage,
       responseMessage: responseMessage,
     };
@@ -151,7 +197,7 @@ const handleAbortError = async (res, req, error, data) => {
       }
     };
 
-    await sendError(res, options, callback);
+    await sendError(req, res, options, callback);
   };
 
   if (partialText && partialText.length > 5) {

api/server/middleware/abortRun.js

@@ -1,6 +1,7 @@
 const { CacheKeys, RunStatus, isUUID } = require('librechat-data-provider');
 const { initializeClient } = require('~/server/services/Endpoints/assistants');
 const { checkMessageGaps, recordUsage } = require('~/server/services/Threads');
+const { deleteMessages } = require('~/models/Message');
 const { getConvo } = require('~/models/Conversation');
 const getLogStores = require('~/cache/getLogStores');
 const { sendMessage } = require('~/server/utils');
@@ -10,7 +11,7 @@ const three_minutes = 1000 * 60 * 3;
 
 async function abortRun(req, res) {
   res.setHeader('Content-Type', 'application/json');
-  const { abortKey } = req.body;
+  const { abortKey, endpoint } = req.body;
   const [conversationId, latestMessageId] = abortKey.split(':');
   const conversation = await getConvo(req.user.id, conversationId);
 
@@ -66,12 +67,19 @@ async function abortRun(req, res) {
     logger.error('[abortRun] Error fetching or processing run', error);
   }
 
+  /* TODO: a reconciling strategy between the existing intermediate message would be more optimal than deleting it */
+  await deleteMessages({
+    user: req.user.id,
+    unfinished: true,
+    conversationId,
+  });
   runMessages = await checkMessageGaps({
     openai,
-    latestMessageId,
-    thread_id,
     run_id,
+    endpoint,
+    thread_id,
     conversationId,
+    latestMessageId,
   });
 
   const finalEvent = {

api/server/middleware/assistants/validate.js

@@ -0,0 +1,44 @@
+const { v4 } = require('uuid');
+const { handleAbortError } = require('~/server/middleware/abortMiddleware');
+
+/**
+ * Checks if the assistant is supported or excluded
+ * @param {object} req - Express Request
+ * @param {object} req.body - The request payload.
+ * @param {object} res - Express Response
+ * @param {function} next - Express next middleware function.
+ * @returns {Promise<void>}
+ */
+const validateAssistant = async (req, res, next) => {
+  const { endpoint, conversationId, assistant_id, messageId } = req.body;
+
+  /** @type {Partial<TAssistantEndpoint>} */
+  const assistantsConfig = req.app.locals?.[endpoint];
+  if (!assistantsConfig) {
+    return next();
+  }
+
+  const { supportedIds, excludedIds } = assistantsConfig;
+  const error = { message: 'validateAssistant: Assistant not supported' };
+
+  if (supportedIds?.length && !supportedIds.includes(assistant_id)) {
+    return await handleAbortError(res, req, error, {
+      sender: 'System',
+      conversationId,
+      messageId: v4(),
+      parentMessageId: messageId,
+      error,
+    });
+  } else if (excludedIds?.length && excludedIds.includes(assistant_id)) {
+    return await handleAbortError(res, req, error, {
+      sender: 'System',
+      conversationId,
+      messageId: v4(),
+      parentMessageId: messageId,
+    });
+  }
+
+  return next();
+};
+
+module.exports = validateAssistant;

api/server/middleware/assistants/validateAuthor.js

@@ -0,0 +1,43 @@
+const { SystemRoles } = require('librechat-data-provider');
+const { getAssistant } = require('~/models/Assistant');
+
+/**
+ * Checks if the assistant is supported or excluded
+ * @param {object} params
+ * @param {object} params.req - Express Request
+ * @param {object} params.req.body - The request payload.
+ * @param {string} params.overrideEndpoint - The override endpoint
+ * @param {string} params.overrideAssistantId - The override assistant ID
+ * @param {OpenAIClient} params.openai - OpenAI API Client
+ * @returns {Promise<void>}
+ */
+const validateAuthor = async ({ req, openai, overrideEndpoint, overrideAssistantId }) => {
+  if (req.user.role === SystemRoles.ADMIN) {
+    return;
+  }
+
+  const endpoint = overrideEndpoint ?? req.body.endpoint ?? req.query.endpoint;
+  const assistant_id =
+    overrideAssistantId ?? req.params.id ?? req.body.assistant_id ?? req.query.assistant_id;
+
+  /** @type {Partial<TAssistantEndpoint>} */
+  const assistantsConfig = req.app.locals?.[endpoint];
+  if (!assistantsConfig) {
+    return;
+  }
+
+  if (!assistantsConfig.privateAssistants) {
+    return;
+  }
+
+  const assistantDoc = await getAssistant({ assistant_id, user: req.user.id });
+  if (assistantDoc) {
+    return;
+  }
+  const assistant = await openai.beta.assistants.retrieve(assistant_id);
+  if (req.user.id !== assistant?.metadata?.author) {
+    throw new Error(`Assistant ${assistant_id} is not authored by the user.`);
+  }
+};
+
+module.exports = validateAuthor;

api/server/middleware/buildEndpointOption.js

@@ -1,5 +1,6 @@
-const { parseConvo, EModelEndpoint } = require('librechat-data-provider');
+const { parseCompactConvo, EModelEndpoint } = require('librechat-data-provider');
 const { getModelsConfig } = require('~/server/controllers/ModelController');
+const azureAssistants = require('~/server/services/Endpoints/azureAssistants');
 const assistants = require('~/server/services/Endpoints/assistants');
 const gptPlugins = require('~/server/services/Endpoints/gptPlugins');
 const { processFiles } = require('~/server/services/Files/process');
@@ -18,11 +19,12 @@ const buildFunction = {
   [EModelEndpoint.anthropic]: anthropic.buildOptions,
   [EModelEndpoint.gptPlugins]: gptPlugins.buildOptions,
   [EModelEndpoint.assistants]: assistants.buildOptions,
+  [EModelEndpoint.azureAssistants]: azureAssistants.buildOptions,
 };
 
 async function buildEndpointOption(req, res, next) {
   const { endpoint, endpointType } = req.body;
-  const parsedBody = parseConvo({ endpoint, endpointType, conversation: req.body });
+  const parsedBody = parseCompactConvo({ endpoint, endpointType, conversation: req.body });
 
   if (req.app.locals.modelSpecs?.list && req.app.locals.modelSpecs?.enforce) {
     /** @type {{ list: TModelSpec[] }}*/