chore: add back Call component from rebase, increase max file size cache

Co-authored-by: github-actions[bot] <41898282+github-actions[bot]@users.noreply.github.com>

.env.example

@@ -53,7 +53,7 @@ DEBUG_CONSOLE=false
 #                     Endpoints                     #
 #===================================================#
 
-# ENDPOINTS=openAI,assistants,azureOpenAI,bingAI,google,gptPlugins,anthropic
+# ENDPOINTS=openAI,assistants,azureOpenAI,google,gptPlugins,anthropic
 
 PROXY=
 
@@ -105,13 +105,6 @@ ANTHROPIC_API_KEY=user_provided
 # AZURE_OPENAI_API_EMBEDDINGS_DEPLOYMENT_NAME= # Deprecated
 # PLUGINS_USE_AZURE="true" # Deprecated
 
-#============#
-# BingAI     #
-#============#
-
-BINGAI_TOKEN=user_provided
-# BINGAI_HOST=https://cn.bing.com
-
 #=================#
 #   AWS Bedrock   #
 #=================#
@@ -177,7 +170,7 @@ GOOGLE_KEY=user_provided
 #============#
 
 OPENAI_API_KEY=user_provided
-# 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
+# OPENAI_MODELS=o1,o1-mini,o1-preview,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
 
@@ -262,6 +255,10 @@ AZURE_AI_SEARCH_SEARCH_OPTION_SELECT=
 GOOGLE_SEARCH_API_KEY=
 GOOGLE_CSE_ID=
 
+# YOUTUBE
+#-----------------
+YOUTUBE_API_KEY=
+
 # SerpAPI
 #-----------------
 SERPAPI_API_KEY=
@@ -392,12 +389,22 @@ FACEBOOK_CALLBACK_URL=/oauth/facebook/callback
 GITHUB_CLIENT_ID=
 GITHUB_CLIENT_SECRET=
 GITHUB_CALLBACK_URL=/oauth/github/callback
+# GitHub Eenterprise
+# GITHUB_ENTERPRISE_BASE_URL=
+# GITHUB_ENTERPRISE_USER_AGENT=
 
 # Google
 GOOGLE_CLIENT_ID=
 GOOGLE_CLIENT_SECRET=
 GOOGLE_CALLBACK_URL=/oauth/google/callback
 
+# Apple
+APPLE_CLIENT_ID=
+APPLE_TEAM_ID=
+APPLE_KEY_ID=
+APPLE_PRIVATE_KEY_PATH=
+APPLE_CALLBACK_URL=/oauth/apple/callback
+
 # OpenID
 OPENID_CLIENT_ID=
 OPENID_CLIENT_SECRET=

.eslintrc.js

@@ -1,213 +0,0 @@
-module.exports = {
-  env: {
-    browser: true,
-    es2021: true,
-    node: true,
-    commonjs: true,
-    es6: true,
-  },
-  extends: [
-    'eslint:recommended',
-    'plugin:react/recommended',
-    'plugin:react-hooks/recommended',
-    'plugin:jest/recommended',
-    'prettier',
-    'plugin:jsx-a11y/recommended',
-  ],
-  ignorePatterns: [
-    'client/dist/**/*',
-    'client/public/**/*',
-    'e2e/playwright-report/**/*',
-    'packages/mcp/types/**/*',
-    'packages/mcp/dist/**/*',
-    'packages/mcp/test_bundle/**/*',
-    'api/demo/**/*',
-    'packages/data-provider/types/**/*',
-    'packages/data-provider/dist/**/*',
-    'packages/data-provider/test_bundle/**/*',
-    'data-node/**/*',
-    'meili_data/**/*',
-    'node_modules/**/*',
-  ],
-  parser: '@typescript-eslint/parser',
-  parserOptions: {
-    ecmaVersion: 'latest',
-    sourceType: 'module',
-    ecmaFeatures: {
-      jsx: true,
-    },
-  },
-  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' }],
-    indent: ['error', 2, { SwitchCase: 1 }],
-    'max-len': [
-      'error',
-      {
-        code: 120,
-        ignoreStrings: true,
-        ignoreTemplateLiterals: true,
-        ignoreComments: true,
-      },
-    ],
-    'linebreak-style': 0,
-    curly: ['error', 'all'],
-    semi: ['error', 'always'],
-    'object-curly-spacing': ['error', 'always'],
-    'no-multiple-empty-lines': ['error', { max: 1 }],
-    'no-trailing-spaces': 'error',
-    'comma-dangle': ['error', 'always-multiline'],
-    // "arrow-parens": [2, "as-needed", { requireForBlockBody: true }],
-    // 'no-plusplus': ['error', { allowForLoopAfterthoughts: true }],
-    'no-console': 'off',
-    'import/no-cycle': 'error',
-    'import/no-self-import': 'error',
-    'import/extensions': 'off',
-    'no-promise-executor-return': 'off',
-    'no-param-reassign': 'off',
-    'no-continue': 'off',
-    'no-restricted-syntax': 'off',
-    'react/prop-types': ['off'],
-    'react/display-name': ['off'],
-    'no-nested-ternary': 'error',
-    'no-unused-vars': ['error', { varsIgnorePattern: '^_' }],
-    quotes: ['error', 'single'],
-  },
-  overrides: [
-    {
-      files: ['**/*.ts', '**/*.tsx'],
-      rules: {
-        'no-unused-vars': 'off', // off because it conflicts with '@typescript-eslint/no-unused-vars'
-        'react/display-name': 'off',
-        '@typescript-eslint/no-unused-vars': 'warn',
-      },
-    },
-    {
-      files: ['rollup.config.js', '.eslintrc.js', 'jest.config.js'],
-      env: {
-        node: true,
-      },
-    },
-    {
-      files: [
-        '**/*.test.js',
-        '**/*.test.jsx',
-        '**/*.test.ts',
-        '**/*.test.tsx',
-        '**/*.spec.js',
-        '**/*.spec.jsx',
-        '**/*.spec.ts',
-        '**/*.spec.tsx',
-        'setupTests.js',
-      ],
-      env: {
-        jest: true,
-        node: true,
-      },
-      rules: {
-        'react/display-name': 'off',
-        'react/prop-types': 'off',
-        'react/no-unescaped-entities': 'off',
-      },
-    },
-    {
-      files: ['**/*.ts', '**/*.tsx'],
-      parser: '@typescript-eslint/parser',
-      parserOptions: {
-        project: './client/tsconfig.json',
-      },
-      plugins: ['@typescript-eslint/eslint-plugin', 'jest'],
-      extends: [
-        'plugin:@typescript-eslint/eslint-recommended',
-        'plugin:@typescript-eslint/recommended',
-      ],
-      rules: {
-        '@typescript-eslint/no-explicit-any': 'error',
-        '@typescript-eslint/no-unnecessary-condition': 'warn',
-        '@typescript-eslint/strict-boolean-expressions': 'warn',
-      },
-    },
-    {
-      files: './packages/data-provider/**/*.ts',
-      overrides: [
-        {
-          files: '**/*.ts',
-          parser: '@typescript-eslint/parser',
-          parserOptions: {
-            project: './packages/data-provider/tsconfig.json',
-          },
-        },
-      ],
-    },
-    {
-      files: './api/demo/**/*.ts',
-      overrides: [
-        {
-          files: '**/*.ts',
-          parser: '@typescript-eslint/parser',
-          parserOptions: {
-            project: './packages/data-provider/tsconfig.json',
-          },
-        },
-      ],
-    },
-    {
-      files: './packages/mcp/**/*.ts',
-      overrides: [
-        {
-          files: '**/*.ts',
-          parser: '@typescript-eslint/parser',
-          parserOptions: {
-            project: './packages/mcp/tsconfig.json',
-          },
-        },
-      ],
-    },
-    {
-      files: './config/translations/**/*.ts',
-      parser: '@typescript-eslint/parser',
-      parserOptions: {
-        project: './config/translations/tsconfig.json',
-      },
-    },
-    {
-      files: ['./packages/data-provider/specs/**/*.ts'],
-      parserOptions: {
-        project: './packages/data-provider/tsconfig.spec.json',
-      },
-    },
-    {
-      files: ['./api/demo/specs/**/*.ts'],
-      parserOptions: {
-        project: './packages/data-provider/tsconfig.spec.json',
-      },
-    },
-    {
-      files: ['./packages/mcp/specs/**/*.ts'],
-      parserOptions: {
-        project: './packages/mcp/tsconfig.spec.json',
-      },
-    },
-  ],
-  settings: {
-    react: {
-      createClass: 'createReactClass', // Regex for Component Factory to use,
-      // default to "createReactClass"
-      pragma: 'React', // Pragma to use, default to "React"
-      fragment: 'Fragment', // Fragment to use (may be a property of <pragma>), default to "Fragment"
-      version: 'detect', // React version. "detect" automatically picks the version you have installed.
-    },
-    'import/parsers': {
-      '@typescript-eslint/parser': ['.ts', '.tsx'],
-    },
-    'import/resolver': {
-      typescript: {
-        project: ['./client/tsconfig.json'],
-      },
-      node: {
-        project: ['./client/tsconfig.json'],
-      },
-    },
-  },
-};

.github/ISSUE_TEMPLATE/BUG-REPORT.yml

@@ -1,12 +1,19 @@
 name: Bug Report
 description: File a bug report
 title: "[Bug]: "
-labels: ["bug"]
+labels: ["🐛 bug"]
 body:
   - type: markdown
     attributes:
       value: |
         Thanks for taking the time to fill out this bug report!
+        
+        Before submitting, please:
+        - Search existing [Issues and Discussions](https://github.com/danny-avila/LibreChat/discussions) to see if your bug has already been reported
+        - Use [Discussions](https://github.com/danny-avila/LibreChat/discussions) instead of Issues for:
+          - General inquiries
+          - Help with setup
+          - Questions about whether you're experiencing a bug
   - type: textarea
     id: what-happened
     attributes:
@@ -15,6 +22,23 @@ body:
       placeholder: Please give as many details as possible
     validations:
       required: true
+  - type: textarea
+    id: version-info
+    attributes:
+      label: Version Information
+      description: |
+        If using Docker, please run and provide the output of:
+        ```bash
+        docker images | grep librechat
+        ```
+        
+        If running from source, please run and provide the output of:
+        ```bash
+        git rev-parse HEAD
+        ```
+      placeholder: Paste the output here
+    validations:
+      required: true
   - type: textarea
     id: steps-to-reproduce
     attributes:
@@ -39,7 +63,21 @@ body:
     id: logs
     attributes:
       label: Relevant log output
-      description: Please copy and paste any relevant log output. This will be automatically formatted into code, so no need for backticks.
+      description: |
+        Please paste relevant logs that were created when reproducing the error.
+        
+        Log locations:
+        - Docker: Project root directory ./logs
+        - npm: ./api/logs
+        
+        There are two types of logs that can help diagnose the issue:
+        - debug logs (debug-YYYY-MM-DD.log)
+        - error logs (error-YYYY-MM-DD.log)
+        
+        Error logs contain exact stack traces and are especially helpful, but both can provide valuable information.
+        Please only include the relevant portions of logs that correspond to when you reproduced the error.
+
+        For UI-related issues, browser console logs can be very helpful. You can provide these as screenshots or paste the text here.
       render: shell
   - type: textarea
     id: screenshots
@@ -53,4 +91,4 @@ body:
       description: By submitting this issue, you agree to follow our [Code of Conduct](https://github.com/danny-avila/LibreChat/blob/main/.github/CODE_OF_CONDUCT.md)
       options:
         - label: I agree to follow this project's Code of Conduct
-          required: true
+          required: true

.github/ISSUE_TEMPLATE/FEATURE-REQUEST.yml

@@ -1,7 +1,7 @@
 name: Feature Request
 description: File a feature request
-title: "Enhancement: "
-labels: ["enhancement"]
+title: "[Enhancement]: "
+labels: ["✨ enhancement"]
 body:
   - type: markdown
     attributes:

.github/ISSUE_TEMPLATE/NEW-LANGUAGE-REQUEST.yml

@@ -0,0 +1,33 @@
+name: New Language Request
+description: Request to add a new language for LibreChat translations.
+title: "New Language Request: "
+labels: ["✨ enhancement", "🌍 i18n"]
+body:
+  - type: markdown
+    attributes:
+      value: |
+        Thank you for taking the time to submit a new language request! Please fill out the following details so we can review your request.
+  - type: input
+    id: language_name
+    attributes:
+      label: Language Name
+      description: Please provide the full name of the language (e.g., Spanish, Mandarin).
+      placeholder: e.g., Spanish
+    validations:
+      required: true
+  - type: input
+    id: iso_code
+    attributes:
+      label: ISO 639-1 Code
+      description: Please provide the ISO 639-1 code for the language (e.g., es for Spanish). You can refer to [this list](https://www.w3schools.com/tags/ref_language_codes.asp) for valid codes.
+      placeholder: e.g., es
+    validations:
+      required: true
+  - type: checkboxes
+    id: terms
+    attributes:
+      label: Code of Conduct
+      description: By submitting this issue, you agree to follow our [Code of Conduct](https://github.com/danny-avila/LibreChat/blob/main/.github/CODE_OF_CONDUCT.md).
+      options:
+        - label: I agree to follow this project's Code of Conduct
+          required: true

.github/ISSUE_TEMPLATE/QUESTION.yml

@@ -1,7 +1,7 @@
 name: Question
 description: Ask your question
 title: "[Question]: "
-labels: ["question"]
+labels: ["❓ question"]
 body:
   - type: markdown
     attributes:

.github/workflows/backend-review.yml

@@ -38,7 +38,7 @@ jobs:
 
       - name: Install MCP Package
         run: npm run build:mcp
-      
+
       - name: Create empty auth.json file
         run: |
           mkdir -p api/data
@@ -61,9 +61,4 @@ jobs:
         run: cd api && npm run test:ci
 
       - name: Run librechat-data-provider unit tests
-        run: cd packages/data-provider && npm run test:ci
-
-      - name: Run linters
-        uses: wearerequired/lint-action@v2
-        with:
-          eslint: true
+        run: cd packages/data-provider && npm run test:ci

.github/workflows/eslint-ci.yml

@@ -0,0 +1,73 @@
+name: ESLint Code Quality Checks
+
+on:
+  pull_request:
+    branches:
+      - main
+      - dev
+      - release/*
+    paths:
+      - 'api/**'
+      - 'client/**'
+
+jobs:
+  eslint_checks:
+    name: Run ESLint Linting
+    runs-on: ubuntu-latest
+    permissions:
+      contents: read
+      security-events: write
+      actions: read
+    steps:
+      - name: Checkout repository
+        uses: actions/checkout@v4
+        with:
+          fetch-depth: 0
+
+      - name: Set up Node.js 20.x
+        uses: actions/setup-node@v4
+        with:
+          node-version: 20
+          cache: npm
+
+      - name: Install dependencies
+        run: npm ci
+
+      # Run ESLint on changed files within the api/ and client/ directories.
+      - name: Run ESLint on changed files
+        env:
+          SARIF_ESLINT_IGNORE_SUPPRESSED: "true"
+        run: |
+          # Extract the base commit SHA from the pull_request event payload.
+          BASE_SHA=$(jq --raw-output .pull_request.base.sha "$GITHUB_EVENT_PATH")
+          echo "Base commit SHA: $BASE_SHA"
+
+          # Get changed files (only JS/TS files in api/ or client/)
+          CHANGED_FILES=$(git diff --name-only --diff-filter=ACMRTUXB "$BASE_SHA" HEAD | grep -E '^(api|client)/.*\.(js|jsx|ts|tsx)$' || true)
+
+          # Debug output
+          echo "Changed files:"
+          echo "$CHANGED_FILES"
+
+          # Ensure there are files to lint before running ESLint
+          if [[ -z "$CHANGED_FILES" ]]; then
+            echo "No matching files changed. Skipping ESLint."
+            echo "UPLOAD_SARIF=false" >> $GITHUB_ENV
+            exit 0
+          fi
+
+          # Set variable to allow SARIF upload
+          echo "UPLOAD_SARIF=true" >> $GITHUB_ENV
+
+          # Run ESLint
+          npx eslint --no-error-on-unmatched-pattern \
+            --config eslint.config.mjs \
+            --format @microsoft/eslint-formatter-sarif \
+            --output-file eslint-results.sarif $CHANGED_FILES || true
+
+      - name: Upload analysis results to GitHub
+        if: env.UPLOAD_SARIF == 'true'
+        uses: github/codeql-action/upload-sarif@v3
+        with:
+          sarif_file: eslint-results.sarif
+          wait-for-processing: true

.github/workflows/i18n-unused-keys.yml

@@ -0,0 +1,84 @@
+name: Detect Unused i18next Strings
+
+on:
+  pull_request:
+    paths:
+      - "client/src/**"
+
+jobs:
+  detect-unused-i18n-keys:
+    runs-on: ubuntu-latest
+    permissions:
+      pull-requests: write  # Required for posting PR comments
+    steps:
+      - name: Checkout repository
+        uses: actions/checkout@v3
+
+      - name: Find unused i18next keys
+        id: find-unused
+        run: |
+          echo "🔍 Scanning for unused i18next keys..."
+
+          # Define paths
+          I18N_FILE="client/src/locales/en/translation.json"
+          SOURCE_DIR="client/src"
+
+          # Check if translation file exists
+          if [[ ! -f "$I18N_FILE" ]]; then
+            echo "::error title=Missing i18n File::Translation file not found: $I18N_FILE"
+            exit 1
+          fi
+
+          # Extract all keys from the JSON file
+          KEYS=$(jq -r 'keys[]' "$I18N_FILE")
+
+          # Track unused keys
+          UNUSED_KEYS=()
+
+          # Check if each key is used in the source code
+          for KEY in $KEYS; do
+            if ! grep -r --include=\*.{js,jsx,ts,tsx} -q "$KEY" "$SOURCE_DIR"; then
+              UNUSED_KEYS+=("$KEY")
+            fi
+          done
+
+          # Output results
+          if [[ ${#UNUSED_KEYS[@]} -gt 0 ]]; then
+            echo "🛑 Found ${#UNUSED_KEYS[@]} unused i18n keys:"
+            echo "unused_keys=$(echo "${UNUSED_KEYS[@]}" | jq -R -s -c 'split(" ")')" >> $GITHUB_ENV
+            for KEY in "${UNUSED_KEYS[@]}"; do
+              echo "::warning title=Unused i18n Key::'$KEY' is defined but not used in the codebase."
+            done
+          else
+            echo "✅ No unused i18n keys detected!"
+            echo "unused_keys=[]" >> $GITHUB_ENV
+          fi
+
+      - name: Post verified comment on PR
+        if: env.unused_keys != '[]'
+        run: |
+          PR_NUMBER=$(jq --raw-output .pull_request.number "$GITHUB_EVENT_PATH")
+
+          # Format the unused keys list as checkboxes for easy manual checking.
+          FILTERED_KEYS=$(echo "$unused_keys" | jq -r '.[]' | grep -v '^\s*$' | sed 's/^/- [ ] `/;s/$/`/' )
+
+          COMMENT_BODY=$(cat <<EOF
+          ### 🚨 Unused i18next Keys Detected
+
+          The following translation keys are defined in \`translation.json\` but are **not used** in the codebase:
+
+          $FILTERED_KEYS
+
+          ⚠️ **Please remove these unused keys to keep the translation files clean.**
+          EOF
+          )
+
+          gh api "repos/${{ github.repository }}/issues/${PR_NUMBER}/comments" \
+            -f body="$COMMENT_BODY" \
+            -H "Authorization: token ${{ secrets.GITHUB_TOKEN }}"
+        env:
+          GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
+
+      - name: Fail workflow if unused keys found
+        if: env.unused_keys != '[]'
+        run: exit 1

.github/workflows/locize-i18n-sync.yml

@@ -0,0 +1,72 @@
+name: Sync Locize Translations & Create Translation PR
+
+on:
+  push:
+    branches: [main]
+  repository_dispatch:
+    types: [locize/versionPublished]
+
+jobs:
+  sync-translations:
+    name: Sync Translation Keys with Locize
+    runs-on: ubuntu-latest
+    steps:
+      - name: Checkout Repository
+        uses: actions/checkout@v4
+
+      - name: Set Up Node.js
+        uses: actions/setup-node@v4
+        with:
+          node-version: 20
+
+      - name: Install locize CLI
+        run: npm install -g locize-cli
+
+      # Sync translations (Push missing keys & remove deleted ones)
+      - name: Sync Locize with Repository
+        if: ${{ github.event_name == 'push' }}
+        run: |
+          cd client/src/locales
+          locize sync --api-key ${{ secrets.LOCIZE_API_KEY }} --project-id ${{ secrets.LOCIZE_PROJECT_ID }} --language en
+
+      # When triggered by repository_dispatch, skip sync step.
+      - name: Skip sync step on non-push events
+        if: ${{ github.event_name != 'push' }}
+        run: echo "Skipping sync as the event is not a push."
+
+  create-pull-request:
+    name: Create Translation PR on Version Published
+    runs-on: ubuntu-latest
+    needs: sync-translations
+    permissions:
+      contents: write
+      pull-requests: write
+    steps:
+      # 1. Check out the repository.
+      - name: Checkout Repository
+        uses: actions/checkout@v4
+
+      # 2. Download translation files from locize.
+      - name: Download Translations from locize
+        uses: locize/download@v1
+        with:
+          project-id: ${{ secrets.LOCIZE_PROJECT_ID }}
+          path: "client/src/locales"
+
+      # 3. Create a Pull Request using built-in functionality.
+      - name: Create Pull Request
+        uses: peter-evans/create-pull-request@v7
+        with:
+          token: ${{ secrets.GITHUB_TOKEN }}
+          sign-commits: true
+          commit-message: "🌍 i18n: Update translation.json with latest translations"
+          base: main
+          branch: i18n/locize-translation-update
+          reviewers: danny-avila
+          title: "🌍 i18n: Update translation.json with latest translations"
+          body: |
+            **Description**:
+            - 🎯 **Objective**: Update `translation.json` with the latest translations from locize.
+            - 🔍 **Details**: This PR is automatically generated upon receiving a versionPublished event with version "latest". It reflects the newest translations provided by locize.
+            - ✅ **Status**: Ready for review.
+          labels: "🌍 i18n"

.github/workflows/unused-packages.yml

@@ -0,0 +1,147 @@
+name: Detect Unused NPM Packages
+
+on: [pull_request]
+
+jobs:
+  detect-unused-packages:
+    runs-on: ubuntu-latest
+    permissions:
+      pull-requests: write
+
+    steps:
+      - uses: actions/checkout@v4
+
+      - name: Use Node.js 20.x
+        uses: actions/setup-node@v4
+        with:
+          node-version: 20
+          cache: 'npm'
+
+      - name: Install depcheck
+        run: npm install -g depcheck
+
+      - name: Validate JSON files
+        run: |
+          for FILE in package.json client/package.json api/package.json; do
+            if [[ -f "$FILE" ]]; then
+              jq empty "$FILE" || (echo "::error title=Invalid JSON::$FILE is invalid" && exit 1)
+            fi
+          done
+
+      - name: Extract Dependencies Used in Scripts
+        id: extract-used-scripts
+        run: |
+          extract_deps_from_scripts() {
+            local package_file=$1
+            if [[ -f "$package_file" ]]; then
+              jq -r '.scripts | to_entries[].value' "$package_file" | \
+                grep -oE '([a-zA-Z0-9_-]+)' | sort -u > used_scripts.txt
+            else
+              touch used_scripts.txt
+            fi
+          }
+
+          extract_deps_from_scripts "package.json"
+          mv used_scripts.txt root_used_deps.txt
+
+          extract_deps_from_scripts "client/package.json"
+          mv used_scripts.txt client_used_deps.txt
+
+          extract_deps_from_scripts "api/package.json"
+          mv used_scripts.txt api_used_deps.txt
+
+      - name: Extract Dependencies Used in Source Code
+        id: extract-used-code
+        run: |
+          extract_deps_from_code() {
+            local folder=$1
+            local output_file=$2
+            if [[ -d "$folder" ]]; then
+              grep -rEho "require\\(['\"]([a-zA-Z0-9@/._-]+)['\"]\\)" "$folder" --include=\*.{js,ts,mjs,cjs} | \
+                sed -E "s/require\\(['\"]([a-zA-Z0-9@/._-]+)['\"]\\)/\1/" > "$output_file"
+          
+              grep -rEho "import .* from ['\"]([a-zA-Z0-9@/._-]+)['\"]" "$folder" --include=\*.{js,ts,mjs,cjs} | \
+                sed -E "s/import .* from ['\"]([a-zA-Z0-9@/._-]+)['\"]/\1/" >> "$output_file"
+          
+              sort -u "$output_file" -o "$output_file"
+            else
+              touch "$output_file"
+            fi
+          }
+
+          extract_deps_from_code "." root_used_code.txt
+          extract_deps_from_code "client" client_used_code.txt
+          extract_deps_from_code "api" api_used_code.txt
+
+      - name: Run depcheck for root package.json
+        id: check-root
+        run: |
+          if [[ -f "package.json" ]]; then
+            UNUSED=$(depcheck --json | jq -r '.dependencies | join("\n")' || echo "")
+            UNUSED=$(comm -23 <(echo "$UNUSED" | sort) <(cat root_used_deps.txt root_used_code.txt | sort) || echo "")
+            echo "ROOT_UNUSED<<EOF" >> $GITHUB_ENV
+            echo "$UNUSED" >> $GITHUB_ENV
+            echo "EOF" >> $GITHUB_ENV
+          fi
+
+      - name: Run depcheck for client/package.json
+        id: check-client
+        run: |
+          if [[ -f "client/package.json" ]]; then
+            chmod -R 755 client
+            cd client
+            UNUSED=$(depcheck --json | jq -r '.dependencies | join("\n")' || echo "")
+            UNUSED=$(comm -23 <(echo "$UNUSED" | sort) <(cat ../client_used_deps.txt ../client_used_code.txt | sort) || echo "")
+            echo "CLIENT_UNUSED<<EOF" >> $GITHUB_ENV
+            echo "$UNUSED" >> $GITHUB_ENV
+            echo "EOF" >> $GITHUB_ENV
+            cd ..
+          fi
+
+      - name: Run depcheck for api/package.json
+        id: check-api
+        run: |
+          if [[ -f "api/package.json" ]]; then
+            chmod -R 755 api
+            cd api
+            UNUSED=$(depcheck --json | jq -r '.dependencies | join("\n")' || echo "")
+            UNUSED=$(comm -23 <(echo "$UNUSED" | sort) <(cat ../api_used_deps.txt ../api_used_code.txt | sort) || echo "")
+            echo "API_UNUSED<<EOF" >> $GITHUB_ENV
+            echo "$UNUSED" >> $GITHUB_ENV
+            echo "EOF" >> $GITHUB_ENV
+            cd ..
+          fi
+
+      - name: Post comment on PR if unused dependencies are found
+        if: env.ROOT_UNUSED != '' || env.CLIENT_UNUSED != '' || env.API_UNUSED != ''
+        run: |
+          PR_NUMBER=$(jq --raw-output .pull_request.number "$GITHUB_EVENT_PATH")
+
+          ROOT_LIST=$(echo "$ROOT_UNUSED" | awk '{print "- `" $0 "`"}')
+          CLIENT_LIST=$(echo "$CLIENT_UNUSED" | awk '{print "- `" $0 "`"}')
+          API_LIST=$(echo "$API_UNUSED" | awk '{print "- `" $0 "`"}')
+
+          COMMENT_BODY=$(cat <<EOF
+          ### 🚨 Unused NPM Packages Detected
+
+          The following **unused dependencies** were found:
+
+          $(if [[ ! -z "$ROOT_UNUSED" ]]; then echo "#### 📂 Root \`package.json\`"; echo ""; echo "$ROOT_LIST"; echo ""; fi)
+
+          $(if [[ ! -z "$CLIENT_UNUSED" ]]; then echo "#### 📂 Client \`client/package.json\`"; echo ""; echo "$CLIENT_LIST"; echo ""; fi)
+
+          $(if [[ ! -z "$API_UNUSED" ]]; then echo "#### 📂 API \`api/package.json\`"; echo ""; echo "$API_LIST"; echo ""; fi)
+
+          ⚠️ **Please remove these unused dependencies to keep your project clean.**
+          EOF
+          )
+
+          gh api "repos/${{ github.repository }}/issues/${PR_NUMBER}/comments" \
+            -f body="$COMMENT_BODY" \
+            -H "Authorization: token ${{ secrets.GITHUB_TOKEN }}"
+        env:
+          GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
+
+      - name: Fail workflow if unused dependencies found
+        if: env.ROOT_UNUSED != '' || env.CLIENT_UNUSED != '' || env.API_UNUSED != ''
+        run: exit 1

.gitignore

@@ -105,4 +105,5 @@ auth.json
 uploads/
 
 # owner
-release/
+release/
+!/client/src/@types/i18next.d.ts

.prettierrc

@@ -0,0 +1,19 @@
+{
+  "tailwindConfig": "./client/tailwind.config.cjs",
+  "printWidth": 100,
+  "tabWidth": 2,
+  "useTabs": false,
+  "semi": true,
+  "singleQuote": true,
+  "trailingComma": "all",
+  "arrowParens": "always",
+  "embeddedLanguageFormatting": "auto",
+  "insertPragma": false,
+  "proseWrap": "preserve",
+  "quoteProps": "as-needed",
+  "requirePragma": false,
+  "rangeStart": 0,
+  "endOfLine": "auto",
+  "jsxSingleQuote": false,
+  "plugins": ["prettier-plugin-tailwindcss"]
+}

Dockerfile

@@ -1,4 +1,4 @@
-# v0.7.6
+# v0.7.7-rc1
 
 # Base node image
 FROM node:20-alpine AS node

Dockerfile.multi

@@ -1,5 +1,5 @@
 # Dockerfile.multi
-# v0.7.6
+# v0.7.7-rc1
 
 # Base for all builds
 FROM node:20-alpine AS base-min

README.md

@@ -38,6 +38,15 @@
 </a>
 </p>
 
+<p align="center">
+  <a href="https://www.librechat.ai/docs/translation">
+    <img 
+      src="https://img.shields.io/badge/dynamic/json.svg?style=for-the-badge&color=2096F3&label=locize&query=%24.translatedPercentage&url=https://api.locize.app/badgedata/4cb2598b-ed4d-469c-9b04-2ed531a8cb45&suffix=%+translated" 
+      alt="Translation Progress">
+  </a>
+</p>
+
+
 # ✨ Features
 
 - 🖥️ **UI & Experience** inspired by ChatGPT with enhanced design and features
@@ -79,6 +88,9 @@
   - 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
 
+- 🧠 **Reasoning UI**:  
+  - Dynamic Reasoning UI for Chain-of-Thought/Reasoning AI models like DeepSeek-R1
+
 - 🎨 **Customizable Interface**:  
   - Customizable Dropdown & Interface that adapts to both power users and newcomers
 
@@ -167,6 +179,8 @@ Contributions, suggestions, bug reports and fixes are welcome!
 
 For new features, components, or extensions, please open an issue and discuss before sending a PR.
 
+If you'd like to help translate LibreChat into your language, we'd love your contribution! Improving our translations not only makes LibreChat more accessible to users around the world but also enhances the overall user experience. Please check out our [Translation Guide](https://www.librechat.ai/docs/translation).
+
 ---
 
 ## 💖 This project exists in its current state thanks to all the people who contribute
@@ -174,3 +188,15 @@ For new features, components, or extensions, please open an issue and discuss be
 <a href="https://github.com/danny-avila/LibreChat/graphs/contributors">
   <img src="https://contrib.rocks/image?repo=danny-avila/LibreChat" />
 </a>
+
+---
+
+## 🎉 Special Thanks
+
+We thank [Locize](https://locize.com) for their translation management tools that support multiple languages in LibreChat.
+
+<p align="center">
+  <a href="https://locize.com" target="_blank" rel="noopener noreferrer">
+    <img src="https://locize.com/img/locize_color.svg" alt="Locize Logo" height="50">
+  </a>
+</p>

api/app/bingai.js

@@ -1,112 +0,0 @@
-require('dotenv').config();
-const { KeyvFile } = require('keyv-file');
-const { EModelEndpoint } = require('librechat-data-provider');
-const { getUserKey, checkUserKeyExpiry } = require('~/server/services/UserService');
-const { logger } = require('~/config');
-
-const askBing = async ({
-  text,
-  parentMessageId,
-  conversationId,
-  jailbreak,
-  jailbreakConversationId,
-  context,
-  systemMessage,
-  conversationSignature,
-  clientId,
-  invocationId,
-  toneStyle,
-  key: expiresAt,
-  onProgress,
-  userId,
-}) => {
-  const isUserProvided = process.env.BINGAI_TOKEN === 'user_provided';
-
-  let key = null;
-  if (expiresAt && isUserProvided) {
-    checkUserKeyExpiry(expiresAt, EModelEndpoint.bingAI);
-    key = await getUserKey({ userId, name: 'bingAI' });
-  }
-
-  const { BingAIClient } = await import('nodejs-gpt');
-  const store = {
-    store: new KeyvFile({ filename: './data/cache.json' }),
-  };
-
-  const bingAIClient = new BingAIClient({
-    // "_U" cookie from bing.com
-    // userToken:
-    //   isUserProvided ? key : process.env.BINGAI_TOKEN ?? null,
-    // If the above doesn't work, provide all your cookies as a string instead
-    cookies: isUserProvided ? key : process.env.BINGAI_TOKEN ?? null,
-    debug: false,
-    cache: store,
-    host: process.env.BINGAI_HOST || null,
-    proxy: process.env.PROXY || null,
-  });
-
-  let options = {};
-
-  if (jailbreakConversationId == 'false') {
-    jailbreakConversationId = false;
-  }
-
-  if (jailbreak) {
-    options = {
-      jailbreakConversationId: jailbreakConversationId || jailbreak,
-      context,
-      systemMessage,
-      parentMessageId,
-      toneStyle,
-      onProgress,
-      clientOptions: {
-        features: {
-          genImage: {
-            server: {
-              enable: true,
-              type: 'markdown_list',
-            },
-          },
-        },
-      },
-    };
-  } else {
-    options = {
-      conversationId,
-      context,
-      systemMessage,
-      parentMessageId,
-      toneStyle,
-      onProgress,
-      clientOptions: {
-        features: {
-          genImage: {
-            server: {
-              enable: true,
-              type: 'markdown_list',
-            },
-          },
-        },
-      },
-    };
-
-    // don't give those parameters for new conversation
-    // for new conversation, conversationSignature always is null
-    if (conversationSignature) {
-      options.encryptedConversationSignature = conversationSignature;
-      options.clientId = clientId;
-      options.invocationId = invocationId;
-    }
-  }
-
-  logger.debug('bing options', options);
-
-  const res = await bingAIClient.sendMessage(text, options);
-
-  return res;
-
-  // for reference:
-  // https://github.com/waylaidwanderer/node-chatgpt-api/blob/main/demos/use-bing-client.js
-};
-
-module.exports = { askBing };

api/app/chatgpt-browser.js

@@ -1,57 +0,0 @@
-require('dotenv').config();
-const { KeyvFile } = require('keyv-file');
-const { Constants, EModelEndpoint } = require('librechat-data-provider');
-const { getUserKey, checkUserKeyExpiry } = require('../server/services/UserService');
-
-const browserClient = async ({
-  text,
-  parentMessageId,
-  conversationId,
-  model,
-  key: expiresAt,
-  onProgress,
-  onEventMessage,
-  abortController,
-  userId,
-}) => {
-  const isUserProvided = process.env.CHATGPT_TOKEN === 'user_provided';
-
-  let key = null;
-  if (expiresAt && isUserProvided) {
-    checkUserKeyExpiry(expiresAt, EModelEndpoint.chatGPTBrowser);
-    key = await getUserKey({ userId, name: 'chatGPTBrowser' });
-  }
-
-  const { ChatGPTBrowserClient } = await import('nodejs-gpt');
-  const store = {
-    store: new KeyvFile({ filename: './data/cache.json' }),
-  };
-
-  const clientOptions = {
-    // Warning: This will expose your access token to a third party. Consider the risks before using this.
-    reverseProxyUrl:
-      process.env.CHATGPT_REVERSE_PROXY ?? 'https://ai.fakeopen.com/api/conversation',
-    // Access token from https://chat.openai.com/api/auth/session
-    accessToken: isUserProvided ? key : process.env.CHATGPT_TOKEN ?? null,
-    model: model,
-    debug: false,
-    proxy: process.env.PROXY ?? null,
-    user: userId,
-  };
-
-  const client = new ChatGPTBrowserClient(clientOptions, store);
-  let options = { onProgress, onEventMessage, abortController };
-
-  if (!!parentMessageId && !!conversationId) {
-    options = { ...options, parentMessageId, conversationId };
-  }
-
-  if (parentMessageId === Constants.NO_PARENT) {
-    delete options.conversationId;
-  }
-
-  const res = await client.sendMessage(text, options);
-  return res;
-};
-
-module.exports = { browserClient };

api/app/clients/AnthropicClient.js

@@ -416,7 +416,7 @@ class AnthropicClient extends BaseClient {
     }
 
     let { context: messagesInWindow, remainingContextTokens } =
-      await this.getMessagesWithinTokenLimit(formattedMessages);
+      await this.getMessagesWithinTokenLimit({ messages: formattedMessages });
 
     const tokenCountMap = orderedMessages
       .slice(orderedMessages.length - messagesInWindow.length)

api/app/clients/BaseClient.js

@@ -7,15 +7,12 @@ const {
   EModelEndpoint,
   ErrorTypes,
   Constants,
-  CacheKeys,
-  Time,
 } = require('librechat-data-provider');
 const { getMessages, saveMessage, updateMessage, saveConvo } = require('~/models');
 const { addSpaceIfNeeded, isEnabled } = require('~/server/utils');
 const { truncateToolCallOutputs } = require('./prompts');
 const checkBalance = require('~/models/checkBalance');
 const { getFiles } = require('~/models/File');
-const { getLogStores } = require('~/cache');
 const TextStream = require('./TextStream');
 const { logger } = require('~/config');
 
@@ -54,6 +51,14 @@ class BaseClient {
     this.outputTokensKey = 'completion_tokens';
     /** @type {Set<string>} */
     this.savedMessageIds = new Set();
+    /**
+     * Flag to determine if the client re-submitted the latest assistant message.
+     * @type {boolean | undefined} */
+    this.continued;
+    /** @type {TMessage[]} */
+    this.currentMessages = [];
+    /** @type {import('librechat-data-provider').VisionModes | undefined} */
+    this.visionMode;
   }
 
   setOptions() {
@@ -264,17 +269,24 @@ class BaseClient {
   /**
    * Adds instructions to the messages array. If the instructions object is empty or undefined,
    * the original messages array is returned. Otherwise, the instructions are added to the messages
-   * array, preserving the last message at the end.
+   * array either at the beginning (default) or preserving the last message at the end.
    *
    * @param {Array} messages - An array of messages.
    * @param {Object} instructions - An object containing instructions to be added to the messages.
+   * @param {boolean} [beforeLast=false] - If true, adds instructions before the last message; if false, adds at the beginning.
    * @returns {Array} An array containing messages and instructions, or the original messages if instructions are empty.
    */
-  addInstructions(messages, instructions) {
-    const payload = [];
+  addInstructions(messages, instructions, beforeLast = false) {
     if (!instructions || Object.keys(instructions).length === 0) {
       return messages;
     }
+
+    if (!beforeLast) {
+      return [instructions, ...messages];
+    }
+
+    // Legacy behavior: add instructions before the last message
+    const payload = [];
     if (messages.length > 1) {
       payload.push(...messages.slice(0, -1));
     }
@@ -340,25 +352,38 @@ class BaseClient {
    * If the token limit would be exceeded by adding a message, that message is not added to the context and remains in the original array.
    * The method uses `push` and `pop` operations for efficient array manipulation, and reverses the context array at the end to maintain the original order of the messages.
    *
-   * @param {Array} _messages - An array of messages, each with a `tokenCount` property. The messages should be ordered from oldest to newest.
-   * @param {number} [maxContextTokens] - The max number of tokens allowed in the context. If not provided, defaults to `this.maxContextTokens`.
-   * @returns {Object} An object with four properties: `context`, `summaryIndex`, `remainingContextTokens`, and `messagesToRefine`.
+   * @param {Object} params
+   * @param {TMessage[]} params.messages - An array of messages, each with a `tokenCount` property. The messages should be ordered from oldest to newest.
+   * @param {number} [params.maxContextTokens] - The max number of tokens allowed in the context. If not provided, defaults to `this.maxContextTokens`.
+   * @param {{ role: 'system', content: text, tokenCount: number }} [params.instructions] - Instructions already added to the context at index 0.
+   * @returns {Promise<{
+   *  context: TMessage[],
+   *  remainingContextTokens: number,
+   *  messagesToRefine: TMessage[],
+   *  summaryIndex: number,
+   * }>} An object with four properties: `context`, `summaryIndex`, `remainingContextTokens`, and `messagesToRefine`.
    *    `context` is an array of messages that fit within the token limit.
    *    `summaryIndex` is the index of the first message in the `messagesToRefine` array.
    *    `remainingContextTokens` is the number of tokens remaining within the limit after adding the messages to the context.
    *    `messagesToRefine` is an array of messages that were not added to the context because they would have exceeded the token limit.
    */
-  async getMessagesWithinTokenLimit(_messages, maxContextTokens) {
+  async getMessagesWithinTokenLimit({ messages: _messages, maxContextTokens, instructions }) {
     // Every reply is primed with <|start|>assistant<|message|>, so we
     // start with 3 tokens for the label after all messages have been counted.
-    let currentTokenCount = 3;
     let summaryIndex = -1;
-    let remainingContextTokens = maxContextTokens ?? this.maxContextTokens;
+    let currentTokenCount = 3;
+    const instructionsTokenCount = instructions?.tokenCount ?? 0;
+    let remainingContextTokens =
+      (maxContextTokens ?? this.maxContextTokens) - instructionsTokenCount;
     const messages = [..._messages];
 
     const context = [];
+
     if (currentTokenCount < remainingContextTokens) {
       while (messages.length > 0 && currentTokenCount < remainingContextTokens) {
+        if (messages.length === 1 && instructions) {
+          break;
+        }
         const poppedMessage = messages.pop();
         const { tokenCount } = poppedMessage;
 
@@ -372,6 +397,11 @@ class BaseClient {
       }
     }
 
+    if (instructions) {
+      context.push(_messages[0]);
+      messages.shift();
+    }
+
     const prunedMemory = messages;
     summaryIndex = prunedMemory.length - 1;
     remainingContextTokens -= currentTokenCount;
@@ -396,12 +426,18 @@ class BaseClient {
     if (instructions) {
       ({ tokenCount, ..._instructions } = instructions);
     }
+
     _instructions && logger.debug('[BaseClient] instructions tokenCount: ' + tokenCount);
-    let payload = this.addInstructions(formattedMessages, _instructions);
-    let orderedWithInstructions = this.addInstructions(orderedMessages, instructions);
+    if (tokenCount && tokenCount > this.maxContextTokens) {
+      const info = `${tokenCount} / ${this.maxContextTokens}`;
+      const errorMessage = `{ "type": "${ErrorTypes.INPUT_LENGTH}", "info": "${info}" }`;
+      logger.warn(`Instructions token count exceeds max token count (${info}).`);
+      throw new Error(errorMessage);
+    }
+
     if (this.clientName === EModelEndpoint.agents) {
       const { dbMessages, editedIndices } = truncateToolCallOutputs(
-        orderedWithInstructions,
+        orderedMessages,
         this.maxContextTokens,
         this.getTokenCountForMessage.bind(this),
       );
@@ -409,14 +445,19 @@ class BaseClient {
       if (editedIndices.length > 0) {
         logger.debug('[BaseClient] Truncated tool call outputs:', editedIndices);
         for (const index of editedIndices) {
-          payload[index].content = dbMessages[index].content;
+          formattedMessages[index].content = dbMessages[index].content;
         }
-        orderedWithInstructions = dbMessages;
+        orderedMessages = dbMessages;
       }
     }
 
+    let orderedWithInstructions = this.addInstructions(orderedMessages, instructions);
+
     let { context, remainingContextTokens, messagesToRefine, summaryIndex } =
-      await this.getMessagesWithinTokenLimit(orderedWithInstructions);
+      await this.getMessagesWithinTokenLimit({
+        messages: orderedWithInstructions,
+        instructions,
+      });
 
     logger.debug('[BaseClient] Context Count (1/2)', {
       remainingContextTokens,
@@ -428,7 +469,9 @@ class BaseClient {
     let { shouldSummarize } = this;
 
     // Calculate the difference in length to determine how many messages were discarded if any
-    const { length } = payload;
+    let payload;
+    let { length } = formattedMessages;
+    length += instructions != null ? 1 : 0;
     const diff = length - context.length;
     const firstMessage = orderedWithInstructions[0];
     const usePrevSummary =
@@ -438,18 +481,31 @@ class BaseClient {
       this.previous_summary.messageId === firstMessage.messageId;
 
     if (diff > 0) {
-      payload = payload.slice(diff);
+      payload = formattedMessages.slice(diff);
       logger.debug(
         `[BaseClient] Difference between original payload (${length}) and context (${context.length}): ${diff}`,
       );
     }
 
+    payload = this.addInstructions(payload ?? formattedMessages, _instructions);
+
     const latestMessage = orderedWithInstructions[orderedWithInstructions.length - 1];
     if (payload.length === 0 && !shouldSummarize && latestMessage) {
       const info = `${latestMessage.tokenCount} / ${this.maxContextTokens}`;
       const errorMessage = `{ "type": "${ErrorTypes.INPUT_LENGTH}", "info": "${info}" }`;
       logger.warn(`Prompt token count exceeds max token count (${info}).`);
       throw new Error(errorMessage);
+    } else if (
+      _instructions &&
+      payload.length === 1 &&
+      payload[0].content === _instructions.content
+    ) {
+      const info = `${tokenCount + 3} / ${this.maxContextTokens}`;
+      const errorMessage = `{ "type": "${ErrorTypes.INPUT_LENGTH}", "info": "${info}" }`;
+      logger.warn(
+        `Including instructions, the prompt token count exceeds remaining max token count (${info}).`,
+      );
+      throw new Error(errorMessage);
     }
 
     if (usePrevSummary) {
@@ -538,6 +594,7 @@ class BaseClient {
       } else {
         latestMessage.text = generation;
       }
+      this.continued = true;
     } else {
       this.currentMessages.push(userMessage);
     }
@@ -669,17 +726,6 @@ class BaseClient {
 
     this.responsePromise = this.saveMessageToDatabase(responseMessage, saveOptions, user);
     this.savedMessageIds.add(responseMessage.messageId);
-    if (responseMessage.text) {
-      const messageCache = getLogStores(CacheKeys.MESSAGES);
-      messageCache.set(
-        responseMessageId,
-        {
-          text: responseMessage.text,
-          complete: true,
-        },
-        Time.FIVE_MINUTES,
-      );
-    }
     delete responseMessage.tokenCount;
     return responseMessage;
   }
@@ -1051,7 +1097,7 @@ class BaseClient {
         file_id: { $in: fileIds },
       });
 
-      await this.addImageURLs(message, files);
+      await this.addImageURLs(message, files, this.visionMode);
 
       this.message_file_map[message.messageId] = files;
       return message;

api/app/clients/ChatGPTClient.js

@@ -13,7 +13,6 @@ const {
 const { extractBaseURL, constructAzureURL, genAzureChatCompletion } = require('~/utils');
 const { createContextHandlers } = require('./prompts');
 const { createCoherePayload } = require('./llm');
-const { Agent, ProxyAgent } = require('undici');
 const BaseClient = require('./BaseClient');
 const { logger } = require('~/config');
 
@@ -186,10 +185,6 @@ class ChatGPTClient extends BaseClient {
       headers: {
         'Content-Type': 'application/json',
       },
-      dispatcher: new Agent({
-        bodyTimeout: 0,
-        headersTimeout: 0,
-      }),
     };
 
     if (this.isVisionModel) {
@@ -275,10 +270,6 @@ class ChatGPTClient extends BaseClient {
       opts.headers['X-Title'] = 'LibreChat';
     }
 
-    if (this.options.proxy) {
-      opts.dispatcher = new ProxyAgent(this.options.proxy);
-    }
-
     /* hacky fixes for Mistral AI API:
       - Re-orders system message to the top of the messages payload, as not allowed anywhere else
       - If there is only one message and it's a system message, change the role to user

api/app/clients/GoogleClient.js

@@ -1,22 +1,25 @@
 const { google } = require('googleapis');
-const { Agent, ProxyAgent } = require('undici');
+const { concat } = require('@langchain/core/utils/stream');
 const { ChatVertexAI } = require('@langchain/google-vertexai');
-const { GoogleVertexAI } = require('@langchain/google-vertexai');
-const { ChatGoogleVertexAI } = require('@langchain/google-vertexai');
 const { ChatGoogleGenerativeAI } = require('@langchain/google-genai');
 const { GoogleGenerativeAI: GenAI } = require('@google/generative-ai');
-const { AIMessage, HumanMessage, SystemMessage } = require('@langchain/core/messages');
+const { HumanMessage, SystemMessage } = require('@langchain/core/messages');
 const {
+  googleGenConfigSchema,
   validateVisionModel,
   getResponseSender,
   endpointSettings,
   EModelEndpoint,
+  ContentTypes,
   VisionModes,
+  ErrorTypes,
   Constants,
   AuthKeys,
 } = require('librechat-data-provider');
+const { getSafetySettings } = require('~/server/services/Endpoints/google/llm');
 const { encodeAndFormat } = require('~/server/services/Files/images');
 const Tokenizer = require('~/server/services/Tokenizer');
+const { spendTokens } = require('~/models/spendTokens');
 const { getModelMaxTokens } = require('~/utils');
 const { sleep } = require('~/server/utils');
 const { logger } = require('~/config');
@@ -49,9 +52,10 @@ class GoogleClient extends BaseClient {
     const serviceKey = creds[AuthKeys.GOOGLE_SERVICE_KEY] ?? {};
     this.serviceKey =
       serviceKey && typeof serviceKey === 'string' ? JSON.parse(serviceKey) : serviceKey ?? {};
+    /** @type {string | null | undefined} */
+    this.project_id = this.serviceKey.project_id;
     this.client_email = this.serviceKey.client_email;
     this.private_key = this.serviceKey.private_key;
-    this.project_id = this.serviceKey.project_id;
     this.access_token = null;
 
     this.apiKey = creds[AuthKeys.GOOGLE_API_KEY];
@@ -60,6 +64,15 @@ class GoogleClient extends BaseClient {
 
     this.authHeader = options.authHeader;
 
+    /** @type {UsageMetadata | undefined} */
+    this.usage;
+    /** The key for the usage object's input tokens
+     * @type {string} */
+    this.inputTokensKey = 'input_tokens';
+    /** The key for the usage object's output tokens
+     * @type {string} */
+    this.outputTokensKey = 'output_tokens';
+    this.visionMode = VisionModes.generative;
     if (options.skipSetOptions) {
       return;
     }
@@ -119,22 +132,13 @@ class GoogleClient extends BaseClient {
       this.options = options;
     }
 
-    this.options.examples = (this.options.examples ?? [])
-      .filter((ex) => ex)
-      .filter((obj) => obj.input.content !== '' && obj.output.content !== '');
-
     this.modelOptions = this.options.modelOptions || {};
 
     this.options.attachments?.then((attachments) => this.checkVisionRequest(attachments));
 
     /** @type {boolean} Whether using a "GenerativeAI" Model */
-    this.isGenerativeModel = this.modelOptions.model.includes('gemini');
-    const { isGenerativeModel } = this;
-    this.isChatModel = !isGenerativeModel && this.modelOptions.model.includes('chat');
-    const { isChatModel } = this;
-    this.isTextModel =
-      !isGenerativeModel && !isChatModel && /code|text/.test(this.modelOptions.model);
-    const { isTextModel } = this;
+    this.isGenerativeModel =
+      this.modelOptions.model.includes('gemini') || this.modelOptions.model.includes('learnlm');
 
     this.maxContextTokens =
       this.options.maxContextTokens ??
@@ -170,40 +174,18 @@ class GoogleClient extends BaseClient {
     this.userLabel = this.options.userLabel || 'User';
     this.modelLabel = this.options.modelLabel || 'Assistant';
 
-    if (isChatModel || isGenerativeModel) {
-      // Use these faux tokens to help the AI understand the context since we are building the chat log ourselves.
-      // Trying to use "<|im_start|>" causes the AI to still generate "<" or "<|" at the end sometimes for some reason,
-      // without tripping the stop sequences, so I'm using "||>" instead.
-      this.startToken = '||>';
-      this.endToken = '';
-    } else if (isTextModel) {
-      this.startToken = '||>';
-      this.endToken = '';
-    } else {
-      // Previously I was trying to use "<|endoftext|>" but there seems to be some bug with OpenAI's token counting
-      // system that causes only the first "<|endoftext|>" to be counted as 1 token, and the rest are not treated
-      // as a single token. So we're using this instead.
-      this.startToken = '||>';
-      this.endToken = '';
-    }
-
-    if (!this.modelOptions.stop) {
-      const stopTokens = [this.startToken];
-      if (this.endToken && this.endToken !== this.startToken) {
-        stopTokens.push(this.endToken);
-      }
-      stopTokens.push(`\n${this.userLabel}:`);
-      stopTokens.push('<|diff_marker|>');
-      // I chose not to do one for `modelLabel` because I've never seen it happen
-      this.modelOptions.stop = stopTokens;
-    }
-
     if (this.options.reverseProxyUrl) {
       this.completionsUrl = this.options.reverseProxyUrl;
     } else {
       this.completionsUrl = this.constructUrl();
     }
 
+    let promptPrefix = (this.options.promptPrefix ?? '').trim();
+    if (typeof this.options.artifactsPrompt === 'string' && this.options.artifactsPrompt) {
+      promptPrefix = `${promptPrefix ?? ''}\n${this.options.artifactsPrompt}`.trim();
+    }
+    this.options.promptPrefix = promptPrefix;
+    this.initializeClient();
     return this;
   }
 
@@ -235,10 +217,29 @@ class GoogleClient extends BaseClient {
   }
 
   formatMessages() {
-    return ((message) => ({
-      author: message?.author ?? (message.isCreatedByUser ? this.userLabel : this.modelLabel),
-      content: message?.content ?? message.text,
-    })).bind(this);
+    return ((message) => {
+      const msg = {
+        author: message?.author ?? (message.isCreatedByUser ? this.userLabel : this.modelLabel),
+        content: message?.content ?? message.text,
+      };
+
+      if (!message.image_urls?.length) {
+        return msg;
+      }
+
+      msg.content = (
+        !Array.isArray(msg.content)
+          ? [
+            {
+              type: ContentTypes.TEXT,
+              [ContentTypes.TEXT]: msg.content,
+            },
+          ]
+          : msg.content
+      ).concat(message.image_urls);
+
+      return msg;
+    }).bind(this);
   }
 
   /**
@@ -336,7 +337,6 @@ class GoogleClient extends BaseClient {
           messages: [new HumanMessage(formatMessage({ message: latestMessage }))],
         },
       ],
-      parameters: this.modelOptions,
     };
     return { prompt: payload };
   }
@@ -352,23 +352,58 @@ class GoogleClient extends BaseClient {
     return { prompt: formattedMessages };
   }
 
-  async buildMessages(messages = [], parentMessageId) {
+  /**
+   * @param {TMessage[]} [messages=[]]
+   * @param {string} [parentMessageId]
+   */
+  async buildMessages(_messages = [], parentMessageId) {
     if (!this.isGenerativeModel && !this.project_id) {
-      throw new Error(
-        '[GoogleClient] a Service Account JSON Key is required for PaLM 2 and Codey models (Vertex AI)',
-      );
+      throw new Error('[GoogleClient] PaLM 2 and Codey models are no longer supported.');
     }
 
+    if (this.options.promptPrefix) {
+      const instructionsTokenCount = this.getTokenCount(this.options.promptPrefix);
+
+      this.maxContextTokens = this.maxContextTokens - instructionsTokenCount;
+      if (this.maxContextTokens < 0) {
+        const info = `${instructionsTokenCount} / ${this.maxContextTokens}`;
+        const errorMessage = `{ "type": "${ErrorTypes.INPUT_LENGTH}", "info": "${info}" }`;
+        logger.warn(`Instructions token count exceeds max context (${info}).`);
+        throw new Error(errorMessage);
+      }
+    }
+
+    for (let i = 0; i < _messages.length; i++) {
+      const message = _messages[i];
+      if (!message.tokenCount) {
+        _messages[i].tokenCount = this.getTokenCountForMessage({
+          role: message.isCreatedByUser ? 'user' : 'assistant',
+          content: message.content ?? message.text,
+        });
+      }
+    }
+
+    const {
+      payload: messages,
+      tokenCountMap,
+      promptTokens,
+    } = await this.handleContextStrategy({
+      orderedMessages: _messages,
+      formattedMessages: _messages,
+    });
+
     if (!this.project_id && !EXCLUDED_GENAI_MODELS.test(this.modelOptions.model)) {
-      return await this.buildGenerativeMessages(messages);
+      const result = await this.buildGenerativeMessages(messages);
+      result.tokenCountMap = tokenCountMap;
+      result.promptTokens = promptTokens;
+      return result;
     }
 
     if (this.options.attachments && this.isGenerativeModel) {
-      return this.buildVisionMessages(messages, parentMessageId);
-    }
-
-    if (this.isTextModel) {
-      return this.buildMessagesPrompt(messages, parentMessageId);
+      const result = this.buildVisionMessages(messages, parentMessageId);
+      result.tokenCountMap = tokenCountMap;
+      result.promptTokens = promptTokens;
+      return result;
     }
 
     let payload = {
@@ -380,25 +415,14 @@ class GoogleClient extends BaseClient {
             .map((message) => formatMessage({ message, langChain: true })),
         },
       ],
-      parameters: this.modelOptions,
     };
 
-    let promptPrefix = (this.options.promptPrefix ?? '').trim();
-    if (typeof this.options.artifactsPrompt === 'string' && this.options.artifactsPrompt) {
-      promptPrefix = `${promptPrefix ?? ''}\n${this.options.artifactsPrompt}`.trim();
-    }
-
-    if (promptPrefix) {
-      payload.instances[0].context = promptPrefix;
-    }
-
-    if (this.options.examples.length > 0) {
-      payload.instances[0].examples = this.options.examples;
+    if (this.options.promptPrefix) {
+      payload.instances[0].context = this.options.promptPrefix;
     }
 
     logger.debug('[GoogleClient] buildMessages', payload);
-
-    return { prompt: payload };
+    return { prompt: payload, tokenCountMap, promptTokens };
   }
 
   async buildMessagesPrompt(messages, parentMessageId) {
@@ -412,10 +436,7 @@ class GoogleClient extends BaseClient {
       parentMessageId,
     });
 
-    const formattedMessages = orderedMessages.map((message) => ({
-      author: message.isCreatedByUser ? this.userLabel : this.modelLabel,
-      content: message?.content ?? message.text,
-    }));
+    const formattedMessages = orderedMessages.map(this.formatMessages());
 
     let lastAuthor = '';
     let groupedMessages = [];
@@ -444,16 +465,6 @@ class GoogleClient extends BaseClient {
     }
 
     let promptPrefix = (this.options.promptPrefix ?? '').trim();
-    if (typeof this.options.artifactsPrompt === 'string' && this.options.artifactsPrompt) {
-      promptPrefix = `${promptPrefix ?? ''}\n${this.options.artifactsPrompt}`.trim();
-    }
-    if (promptPrefix) {
-      // If the prompt prefix doesn't end with the end token, add it.
-      if (!promptPrefix.endsWith(`${this.endToken}`)) {
-        promptPrefix = `${promptPrefix.trim()}${this.endToken}\n\n`;
-      }
-      promptPrefix = `\nContext:\n${promptPrefix}`;
-    }
 
     if (identityPrefix) {
       promptPrefix = `${identityPrefix}${promptPrefix}`;
@@ -490,7 +501,7 @@ class GoogleClient extends BaseClient {
           isCreatedByUser || !isEdited
             ? `\n\n${message.author}:`
             : `${promptPrefix}\n\n${message.author}:`;
-        const messageString = `${messagePrefix}\n${message.content}${this.endToken}\n`;
+        const messageString = `${messagePrefix}\n${message.content}\n`;
         let newPromptBody = `${messageString}${promptBody}`;
 
         context.unshift(message);
@@ -556,34 +567,6 @@ class GoogleClient extends BaseClient {
     return { prompt, context };
   }
 
-  async _getCompletion(payload, abortController = null) {
-    if (!abortController) {
-      abortController = new AbortController();
-    }
-    const { debug } = this.options;
-    const url = this.completionsUrl;
-    if (debug) {
-      logger.debug('GoogleClient _getCompletion', { url, payload });
-    }
-    const opts = {
-      method: 'POST',
-      agent: new Agent({
-        bodyTimeout: 0,
-        headersTimeout: 0,
-      }),
-      signal: abortController.signal,
-    };
-
-    if (this.options.proxy) {
-      opts.agent = new ProxyAgent(this.options.proxy);
-    }
-
-    const client = await this.getClient();
-    const res = await client.request({ url, method: 'POST', data: payload });
-    logger.debug('GoogleClient _getCompletion', { res });
-    return res.data;
-  }
-
   createLLM(clientOptions) {
     const model = clientOptions.modelName ?? clientOptions.model;
     clientOptions.location = loc;
@@ -602,33 +585,30 @@ class GoogleClient extends BaseClient {
       }
     }
 
-    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) {
+    if (this.project_id != null) {
       logger.debug('Creating VertexAI client');
-      return new ChatVertexAI(clientOptions);
+      this.visionMode = undefined;
+      clientOptions.streaming = true;
+      const client = new ChatVertexAI(clientOptions);
+      client.temperature = clientOptions.temperature;
+      client.topP = clientOptions.topP;
+      client.topK = clientOptions.topK;
+      client.topLogprobs = clientOptions.topLogprobs;
+      client.frequencyPenalty = clientOptions.frequencyPenalty;
+      client.presencePenalty = clientOptions.presencePenalty;
+      client.maxOutputTokens = clientOptions.maxOutputTokens;
+      return client;
     } else if (!EXCLUDED_GENAI_MODELS.test(model)) {
       logger.debug('Creating GenAI client');
-      return new GenAI(this.apiKey).getGenerativeModel({ ...clientOptions, model }, requestOptions);
+      return new GenAI(this.apiKey).getGenerativeModel({ model }, requestOptions);
     }
 
     logger.debug('Creating Chat Google Generative AI client');
     return new ChatGoogleGenerativeAI({ ...clientOptions, apiKey: this.apiKey });
   }
 
-  async getCompletion(_payload, 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;
-
-    let clientOptions = { ...parameters, maxRetries: 2 };
+  initializeClient() {
+    let clientOptions = { ...this.modelOptions };
 
     if (this.project_id) {
       clientOptions['authOptions'] = {
@@ -639,184 +619,238 @@ class GoogleClient extends BaseClient {
       };
     }
 
-    if (!parameters) {
-      clientOptions = { ...clientOptions, ...this.modelOptions };
-    }
-
     if (this.isGenerativeModel && !this.project_id) {
       clientOptions.modelName = clientOptions.model;
       delete clientOptions.model;
     }
 
-    if (_examples && _examples.length) {
-      examples = _examples
-        .map((ex) => {
-          const { input, output } = ex;
-          if (!input || !output) {
-            return undefined;
-          }
-          return {
-            input: new HumanMessage(input.content),
-            output: new AIMessage(output.content),
-          };
-        })
-        .filter((ex) => ex);
+    this.client = this.createLLM(clientOptions);
+    return this.client;
+  }
 
-      clientOptions.examples = examples;
-    }
-
-    const model = this.createLLM(clientOptions);
+  async getCompletion(_payload, options = {}) {
+    const { onProgress, abortController } = options;
+    const safetySettings = getSafetySettings(this.modelOptions.model);
+    const streamRate = this.options.streamRate ?? Constants.DEFAULT_STREAM_RATE;
+    const modelName = this.modelOptions.modelName ?? this.modelOptions.model ?? '';
 
     let reply = '';
-    const messages = this.isTextModel ? _payload.trim() : _messages;
-
-    if (!this.isVisionModel && context && messages?.length > 0) {
-      messages.unshift(new SystemMessage(context));
-    }
-
-    const modelName = clientOptions.modelName ?? clientOptions.model ?? '';
-    if (!EXCLUDED_GENAI_MODELS.test(modelName) && !this.project_id) {
-      const client = model;
-      const requestOptions = {
-        contents: _payload,
-      };
-
-      let promptPrefix = (this.options.promptPrefix ?? '').trim();
-      if (typeof this.options.artifactsPrompt === 'string' && this.options.artifactsPrompt) {
-        promptPrefix = `${promptPrefix ?? ''}\n${this.options.artifactsPrompt}`.trim();
-      }
-
-      if (promptPrefix.length) {
-        requestOptions.systemInstruction = {
-          parts: [
-            {
-              text: promptPrefix,
-            },
-          ],
+    /** @type {Error} */
+    let error;
+    try {
+      if (!EXCLUDED_GENAI_MODELS.test(modelName) && !this.project_id) {
+        /** @type {GenAI} */
+        const client = this.client;
+        /** @type {GenerateContentRequest} */
+        const requestOptions = {
+          safetySettings,
+          contents: _payload,
+          generationConfig: googleGenConfigSchema.parse(this.modelOptions),
         };
+
+        const promptPrefix = (this.options.promptPrefix ?? '').trim();
+        if (promptPrefix.length) {
+          requestOptions.systemInstruction = {
+            parts: [
+              {
+                text: promptPrefix,
+              },
+            ],
+          };
+        }
+
+        const delay = modelName.includes('flash') ? 8 : 15;
+        /** @type {GenAIUsageMetadata} */
+        let usageMetadata;
+
+        const result = await client.generateContentStream(requestOptions);
+        for await (const chunk of result.stream) {
+          usageMetadata = !usageMetadata
+            ? chunk?.usageMetadata
+            : Object.assign(usageMetadata, chunk?.usageMetadata);
+          const chunkText = chunk.text();
+          await this.generateTextStream(chunkText, onProgress, {
+            delay,
+          });
+          reply += chunkText;
+          await sleep(streamRate);
+        }
+
+        if (usageMetadata) {
+          this.usage = {
+            input_tokens: usageMetadata.promptTokenCount,
+            output_tokens: usageMetadata.candidatesTokenCount,
+          };
+        }
+
+        return reply;
       }
 
-      requestOptions.safetySettings = _payload.safetySettings;
+      const { instances } = _payload;
+      const { messages: messages, context } = instances?.[0] ?? {};
 
-      const delay = modelName.includes('flash') ? 8 : 15;
-      const result = await client.generateContentStream(requestOptions);
-      for await (const chunk of result.stream) {
-        const chunkText = chunk.text();
+      if (!this.isVisionModel && context && messages?.length > 0) {
+        messages.unshift(new SystemMessage(context));
+      }
+
+      /** @type {import('@langchain/core/messages').AIMessageChunk['usage_metadata']} */
+      let usageMetadata;
+      /** @type {ChatVertexAI} */
+      const client = this.client;
+      const stream = await client.stream(messages, {
+        signal: abortController.signal,
+        streamUsage: true,
+        safetySettings,
+      });
+
+      let delay = this.options.streamRate || 8;
+
+      if (!this.options.streamRate) {
+        if (this.isGenerativeModel) {
+          delay = 15;
+        }
+        if (modelName.includes('flash')) {
+          delay = 5;
+        }
+      }
+
+      for await (const chunk of stream) {
+        if (chunk?.usage_metadata) {
+          const metadata = chunk.usage_metadata;
+          for (const key in metadata) {
+            if (Number.isNaN(metadata[key])) {
+              delete metadata[key];
+            }
+          }
+
+          usageMetadata = !usageMetadata ? metadata : concat(usageMetadata, metadata);
+        }
+
+        const chunkText = chunk?.content ?? '';
         await this.generateTextStream(chunkText, onProgress, {
           delay,
         });
         reply += chunkText;
-        await sleep(streamRate);
       }
-      return reply;
+
+      if (usageMetadata) {
+        this.usage = usageMetadata;
+      }
+    } catch (e) {
+      error = e;
+      logger.error('[GoogleClient] There was an issue generating the completion', e);
     }
 
-    const stream = await model.stream(messages, {
-      signal: abortController.signal,
-      safetySettings: _payload.safetySettings,
-    });
-
-    let delay = this.options.streamRate || 8;
-
-    if (!this.options.streamRate) {
-      if (this.isGenerativeModel) {
-        delay = 15;
-      }
-      if (modelName.includes('flash')) {
-        delay = 5;
-      }
+    if (error != null && reply === '') {
+      const errorMessage = `{ "type": "${ErrorTypes.GoogleError}", "info": "${
+        error.message ?? 'The Google provider failed to generate content, please contact the Admin.'
+      }" }`;
+      throw new Error(errorMessage);
     }
-
-    for await (const chunk of stream) {
-      const chunkText = chunk?.content ?? chunk;
-      await this.generateTextStream(chunkText, onProgress, {
-        delay,
-      });
-      reply += chunkText;
-    }
-
     return reply;
   }
 
+  /**
+   * Get stream usage as returned by this client's API response.
+   * @returns {UsageMetadata} The stream usage object.
+   */
+  getStreamUsage() {
+    return this.usage;
+  }
+
+  /**
+   * Calculates the correct token count for the current user 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 {UsageMetadata} params.usage - The usage object returned by the API.
+   * @returns {number} The correct token count for the current user 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;
+    const currentMessageTokens = totalInputTokens - totalTokensFromMap;
+    return currentMessageTokens > 0 ? currentMessageTokens : originalEstimate;
+  }
+
+  /**
+   * @param {object} params
+   * @param {number} params.promptTokens
+   * @param {number} params.completionTokens
+   * @param {UsageMetadata} [params.usage]
+   * @param {string} [params.model]
+   * @param {string} [params.context='message']
+   * @returns {Promise<void>}
+   */
+  async recordTokenUsage({ promptTokens, completionTokens, model, context = 'message' }) {
+    await spendTokens(
+      {
+        context,
+        user: this.user ?? this.options.req?.user?.id,
+        conversationId: this.conversationId,
+        model: model ?? this.modelOptions.model,
+        endpointTokenConfig: this.options.endpointTokenConfig,
+      },
+      { promptTokens, completionTokens },
+    );
+  }
+
   /**
    * 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 { abortController } = options;
 
-    const modelName = clientOptions.modelName ?? clientOptions.model ?? '';
-    if (!EXCLUDED_GENAI_MODELS.test(modelName) && !this.project_id) {
+    const model = this.modelOptions.modelName ?? this.modelOptions.model ?? '';
+    const safetySettings = getSafetySettings(model);
+    if (!EXCLUDED_GENAI_MODELS.test(model) && !this.project_id) {
       logger.debug('Identified titling model as GenAI version');
       /** @type {GenerativeModel} */
-      const client = model;
+      const client = this.client;
       const requestOptions = {
         contents: _payload,
+        safetySettings,
+        generationConfig: {
+          temperature: 0.5,
+        },
       };
 
-      let promptPrefix = (this.options.promptPrefix ?? '').trim();
-      if (typeof this.options.artifactsPrompt === 'string' && this.options.artifactsPrompt) {
-        promptPrefix = `${promptPrefix ?? ''}\n${this.options.artifactsPrompt}`.trim();
-      }
-
-      if (this.options?.promptPrefix?.length) {
-        requestOptions.systemInstruction = {
-          parts: [
-            {
-              text: promptPrefix,
-            },
-          ],
-        };
-      }
-
-      const safetySettings = _payload.safetySettings;
-      requestOptions.safetySettings = safetySettings;
-
       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, {
+      const { instances } = _payload;
+      const { messages } = instances?.[0] ?? {};
+      const titleResponse = await this.client.invoke(messages, {
         signal: abortController.signal,
         timeout: 7000,
-        safetySettings: safetySettings,
+        safetySettings,
       });
 
+      if (titleResponse.usage_metadata) {
+        await this.recordTokenUsage({
+          model,
+          promptTokens: titleResponse.usage_metadata.input_tokens,
+          completionTokens: titleResponse.usage_metadata.output_tokens,
+          context: 'title',
+        });
+      }
+
       reply = titleResponse.content;
-      // TODO: RECORD TOKEN USAGE
       return reply;
     }
   }
@@ -840,15 +874,8 @@ class GoogleClient extends BaseClient {
       },
     ]);
 
-    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 };
-    }
-
     try {
+      this.initializeClient();
       title = await this.titleChatCompletion(payload, {
         abortController: new AbortController(),
         onProgress: () => {},
@@ -865,6 +892,7 @@ class GoogleClient extends BaseClient {
       endpointType: null,
       artifacts: this.options.artifacts,
       promptPrefix: this.options.promptPrefix,
+      maxContextTokens: this.options.maxContextTokens,
       modelLabel: this.options.modelLabel,
       iconURL: this.options.iconURL,
       greeting: this.options.greeting,
@@ -878,58 +906,31 @@ class GoogleClient extends BaseClient {
   }
 
   async sendCompletion(payload, opts = {}) {
-    payload.safetySettings = this.getSafetySettings();
-
     let reply = '';
     reply = await this.getCompletion(payload, opts);
     return reply.trim();
   }
 
-  getSafetySettings() {
-    const isGemini2 = this.modelOptions.model.includes('gemini-2.0');
-    const mapThreshold = (value) => {
-      if (isGemini2 && value === 'BLOCK_NONE') {
-        return 'OFF';
-      }
-      return value;
-    };
-
-    return [
-      {
-        category: 'HARM_CATEGORY_SEXUALLY_EXPLICIT',
-        threshold: mapThreshold(
-          process.env.GOOGLE_SAFETY_SEXUALLY_EXPLICIT || 'HARM_BLOCK_THRESHOLD_UNSPECIFIED',
-        ),
-      },
-      {
-        category: 'HARM_CATEGORY_HATE_SPEECH',
-        threshold: mapThreshold(
-          process.env.GOOGLE_SAFETY_HATE_SPEECH || 'HARM_BLOCK_THRESHOLD_UNSPECIFIED',
-        ),
-      },
-      {
-        category: 'HARM_CATEGORY_HARASSMENT',
-        threshold: mapThreshold(
-          process.env.GOOGLE_SAFETY_HARASSMENT || 'HARM_BLOCK_THRESHOLD_UNSPECIFIED',
-        ),
-      },
-      {
-        category: 'HARM_CATEGORY_DANGEROUS_CONTENT',
-        threshold: mapThreshold(
-          process.env.GOOGLE_SAFETY_DANGEROUS_CONTENT || 'HARM_BLOCK_THRESHOLD_UNSPECIFIED',
-        ),
-      },
-      {
-        category: 'HARM_CATEGORY_CIVIC_INTEGRITY',
-        threshold: mapThreshold(process.env.GOOGLE_SAFETY_CIVIC_INTEGRITY || 'BLOCK_NONE'),
-      },
-    ];
-  }
-
   getEncoding() {
     return 'cl100k_base';
   }
 
+  async getVertexTokenCount(text) {
+    /** @type {ChatVertexAI} */
+    const client = this.client ?? this.initializeClient();
+    const connection = client.connection;
+    const gAuthClient = connection.client;
+    const tokenEndpoint = `https://${connection._endpoint}/${connection.apiVersion}/projects/${this.project_id}/locations/${connection._location}/publishers/google/models/${connection.model}/:countTokens`;
+    const result = await gAuthClient.request({
+      url: tokenEndpoint,
+      method: 'POST',
+      data: {
+        contents: [{ role: 'user', parts: [{ text }] }],
+      },
+    });
+    return result;
+  }
+
   /**
    * Returns the token count of a given text. It also checks and resets the tokenizers if necessary.
    * @param {string} text - The text to get the token count for.

api/app/clients/OpenAIClient.js

@@ -1,6 +1,7 @@
 const OpenAI = require('openai');
 const { OllamaClient } = require('./OllamaClient');
 const { HttpsProxyAgent } = require('https-proxy-agent');
+const { SplitStreamHandler, GraphEvents } = require('@librechat/agents');
 const {
   Constants,
   ImageDetail,
@@ -28,17 +29,17 @@ const {
   createContextHandlers,
 } = require('./prompts');
 const { encodeAndFormat } = require('~/server/services/Files/images/encode');
+const { addSpaceIfNeeded, isEnabled, sleep } = require('~/server/utils');
 const Tokenizer = require('~/server/services/Tokenizer');
 const { spendTokens } = require('~/models/spendTokens');
-const { isEnabled, sleep } = require('~/server/utils');
 const { handleOpenAIErrors } = require('./tools/util');
 const { createLLM, RunManager } = require('./llm');
+const { logger, sendEvent } = require('~/config');
 const ChatGPTClient = require('./ChatGPTClient');
 const { summaryBuffer } = require('./memory');
 const { runTitleChain } = require('./chains');
 const { tokenSplit } = require('./document');
 const BaseClient = require('./BaseClient');
-const { logger } = require('~/config');
 
 class OpenAIClient extends BaseClient {
   constructor(apiKey, options = {}) {
@@ -64,7 +65,9 @@ class OpenAIClient extends BaseClient {
     /** @type {OpenAIUsageMetadata | undefined} */
     this.usage;
     /** @type {boolean|undefined} */
-    this.isO1Model;
+    this.isOmni;
+    /** @type {SplitStreamHandler | undefined} */
+    this.streamHandler;
   }
 
   // TODO: PluginsClient calls this 3x, unneeded
@@ -102,8 +105,8 @@ class OpenAIClient extends BaseClient {
       this.checkVisionRequest(this.options.attachments);
     }
 
-    const o1Pattern = /\bo1\b/i;
-    this.isO1Model = o1Pattern.test(this.modelOptions.model);
+    const omniPattern = /\b(o1|o3)\b/i;
+    this.isOmni = omniPattern.test(this.modelOptions.model);
 
     const { OPENROUTER_API_KEY, OPENAI_FORCE_PROMPT } = process.env ?? {};
     if (OPENROUTER_API_KEY && !this.azure) {
@@ -143,7 +146,7 @@ class OpenAIClient extends BaseClient {
     const { model } = this.modelOptions;
 
     this.isChatCompletion =
-      o1Pattern.test(model) || model.includes('gpt') || this.useOpenRouter || !!reverseProxy;
+      omniPattern.test(model) || model.includes('gpt') || this.useOpenRouter || !!reverseProxy;
     this.isChatGptModel = this.isChatCompletion;
     if (
       model.includes('text-davinci') ||
@@ -472,7 +475,7 @@ class OpenAIClient extends BaseClient {
       promptPrefix = this.augmentedPrompt + promptPrefix;
     }
 
-    if (promptPrefix && this.isO1Model !== true) {
+    if (promptPrefix && this.isOmni !== true) {
       promptPrefix = `Instructions:\n${promptPrefix.trim()}`;
       instructions = {
         role: 'system',
@@ -500,12 +503,11 @@ class OpenAIClient extends BaseClient {
     };
 
     /** EXPERIMENTAL */
-    if (promptPrefix && this.isO1Model === true) {
+    if (promptPrefix && this.isOmni === true) {
       const lastUserMessageIndex = payload.findLastIndex((message) => message.role === 'user');
       if (lastUserMessageIndex !== -1) {
-        payload[
-          lastUserMessageIndex
-        ].content = `${promptPrefix}\n${payload[lastUserMessageIndex].content}`;
+        payload[lastUserMessageIndex].content =
+          `${promptPrefix}\n${payload[lastUserMessageIndex].content}`;
       }
     }
 
@@ -614,8 +616,6 @@ class OpenAIClient extends BaseClient {
     model = 'gpt-4o-mini',
     modelName,
     temperature = 0.2,
-    presence_penalty = 0,
-    frequency_penalty = 0,
     max_tokens,
     streaming,
     context,
@@ -626,8 +626,6 @@ class OpenAIClient extends BaseClient {
     const modelOptions = {
       modelName: modelName ?? model,
       temperature,
-      presence_penalty,
-      frequency_penalty,
       user: this.user,
     };
 
@@ -798,7 +796,11 @@ ${convo}
         }
 
         title = (
-          await this.sendPayload(instructionsPayload, { modelOptions, useChatCompletion })
+          await this.sendPayload(instructionsPayload, {
+            modelOptions,
+            useChatCompletion,
+            context: 'title',
+          })
         ).replaceAll('"', '');
 
         const completionTokens = this.getTokenCount(title);
@@ -931,7 +933,10 @@ ${convo}
     );
 
     if (excessTokenCount > maxContextTokens) {
-      ({ context } = await this.getMessagesWithinTokenLimit(context, maxContextTokens));
+      ({ context } = await this.getMessagesWithinTokenLimit({
+        messages: context,
+        maxContextTokens,
+      }));
     }
 
     if (context.length === 0) {
@@ -1061,10 +1066,58 @@ ${convo}
     });
   }
 
+  /**
+   *
+   * @param {string[]} [intermediateReply]
+   * @returns {string}
+   */
+  getStreamText(intermediateReply) {
+    if (!this.streamHandler) {
+      return intermediateReply?.join('') ?? '';
+    }
+
+    let thinkMatch;
+    let remainingText;
+    let reasoningText = '';
+
+    if (this.streamHandler.reasoningTokens.length > 0) {
+      reasoningText = this.streamHandler.reasoningTokens.join('');
+      thinkMatch = reasoningText.match(/<think>([\s\S]*?)<\/think>/)?.[1]?.trim();
+      if (thinkMatch != null && thinkMatch) {
+        const reasoningTokens = `:::thinking\n${thinkMatch}\n:::\n`;
+        remainingText = reasoningText.split(/<\/think>/)?.[1]?.trim() || '';
+        return `${reasoningTokens}${remainingText}${this.streamHandler.tokens.join('')}`;
+      } else if (thinkMatch === '') {
+        remainingText = reasoningText.split(/<\/think>/)?.[1]?.trim() || '';
+        return `${remainingText}${this.streamHandler.tokens.join('')}`;
+      }
+    }
+
+    const reasoningTokens =
+      reasoningText.length > 0
+        ? `:::thinking\n${reasoningText.replace('<think>', '').replace('</think>', '').trim()}\n:::\n`
+        : '';
+
+    return `${reasoningTokens}${this.streamHandler.tokens.join('')}`;
+  }
+
+  getMessageMapMethod() {
+    /**
+     * @param {TMessage} msg
+     */
+    return (msg) => {
+      if (msg.text != null && msg.text && msg.text.startsWith(':::thinking')) {
+        msg.text = msg.text.replace(/:::thinking.*?:::/gs, '').trim();
+      }
+
+      return msg;
+    };
+  }
+
   async chatCompletion({ payload, onProgress, abortController = null }) {
     let error = null;
+    let intermediateReply = [];
     const errorCallback = (err) => (error = err);
-    const intermediateReply = [];
     try {
       if (!abortController) {
         abortController = new AbortController();
@@ -1168,7 +1221,7 @@ ${convo}
         opts.defaultHeaders = { ...opts.defaultHeaders, 'api-key': this.apiKey };
       }
 
-      if (this.isO1Model === true && modelOptions.max_tokens != null) {
+      if (this.isOmni === true && modelOptions.max_tokens != null) {
         modelOptions.max_completion_tokens = modelOptions.max_tokens;
         delete modelOptions.max_tokens;
       }
@@ -1248,23 +1301,53 @@ ${convo}
       let streamResolve;
 
       if (
-        this.isO1Model === true &&
+        this.isOmni === true &&
         (this.azure || /o1(?!-(?:mini|preview)).*$/.test(modelOptions.model)) &&
+        !/o3-.*$/.test(this.modelOptions.model) &&
         modelOptions.stream
       ) {
         delete modelOptions.stream;
         delete modelOptions.stop;
+      } else if (!this.isOmni && modelOptions.reasoning_effort != null) {
+        delete modelOptions.reasoning_effort;
       }
 
+      let reasoningKey = 'reasoning_content';
+      if (this.useOpenRouter) {
+        modelOptions.include_reasoning = true;
+        reasoningKey = 'reasoning';
+      }
+
+      this.streamHandler = new SplitStreamHandler({
+        reasoningKey,
+        accumulate: true,
+        runId: this.responseMessageId,
+        handlers: {
+          [GraphEvents.ON_RUN_STEP]: (event) => sendEvent(this.options.res, event),
+          [GraphEvents.ON_MESSAGE_DELTA]: (event) => sendEvent(this.options.res, event),
+          [GraphEvents.ON_REASONING_DELTA]: (event) => sendEvent(this.options.res, event),
+        },
+      });
+
+      intermediateReply = this.streamHandler.tokens;
+
       if (modelOptions.stream) {
         streamPromise = new Promise((resolve) => {
           streamResolve = resolve;
         });
+        /** @type {OpenAI.OpenAI.CompletionCreateParamsStreaming} */
+        const params = {
+          ...modelOptions,
+          stream: true,
+        };
+        if (
+          this.options.endpoint === EModelEndpoint.openAI ||
+          this.options.endpoint === EModelEndpoint.azureOpenAI
+        ) {
+          params.stream_options = { include_usage: true };
+        }
         const stream = await openai.beta.chat.completions
-          .stream({
-            ...modelOptions,
-            stream: true,
-          })
+          .stream(params)
           .on('abort', () => {
             /* Do nothing here */
           })
@@ -1282,20 +1365,44 @@ ${convo}
             }
 
             if (typeof finalMessage.content !== 'string' || finalMessage.content.trim() === '') {
-              finalChatCompletion.choices[0].message.content = intermediateReply.join('');
+              finalChatCompletion.choices[0].message.content = this.streamHandler.tokens.join('');
             }
           })
           .on('finalMessage', (message) => {
             if (message?.role !== 'assistant') {
-              stream.messages.push({ role: 'assistant', content: intermediateReply.join('') });
+              stream.messages.push({
+                role: 'assistant',
+                content: this.streamHandler.tokens.join(''),
+              });
               UnexpectedRoleError = true;
             }
           });
 
+        if (this.continued === true) {
+          const latestText = addSpaceIfNeeded(
+            this.currentMessages[this.currentMessages.length - 1]?.text ?? '',
+          );
+          this.streamHandler.handle({
+            choices: [
+              {
+                delta: {
+                  content: latestText,
+                },
+              },
+            ],
+          });
+        }
+
         for await (const chunk of stream) {
-          const token = chunk?.choices?.[0]?.delta?.content || '';
-          intermediateReply.push(token);
-          onProgress(token);
+          // Add finish_reason: null if missing in any choice
+          if (chunk.choices) {
+            chunk.choices.forEach((choice) => {
+              if (!('finish_reason' in choice)) {
+                choice.finish_reason = null;
+              }
+            });
+          }
+          this.streamHandler.handle(chunk);
           if (abortController.signal.aborted) {
             stream.controller.abort();
             break;
@@ -1338,7 +1445,7 @@ ${convo}
 
       if (!Array.isArray(choices) || choices.length === 0) {
         logger.warn('[OpenAIClient] Chat completion response has no choices');
-        return intermediateReply.join('');
+        return this.streamHandler.tokens.join('');
       }
 
       const { message, finish_reason } = choices[0] ?? {};
@@ -1348,11 +1455,11 @@ ${convo}
 
       if (!message) {
         logger.warn('[OpenAIClient] Message is undefined in chatCompletion response');
-        return intermediateReply.join('');
+        return this.streamHandler.tokens.join('');
       }
 
       if (typeof message.content !== 'string' || message.content.trim() === '') {
-        const reply = intermediateReply.join('');
+        const reply = this.streamHandler.tokens.join('');
         logger.debug(
           '[OpenAIClient] chatCompletion: using intermediateReply due to empty message.content',
           { intermediateReply: reply },
@@ -1360,13 +1467,27 @@ ${convo}
         return reply;
       }
 
+      if (
+        this.streamHandler.reasoningTokens.length > 0 &&
+        this.options.context !== 'title' &&
+        !message.content.startsWith('<think>')
+      ) {
+        return this.getStreamText();
+      } else if (
+        this.streamHandler.reasoningTokens.length > 0 &&
+        this.options.context !== 'title' &&
+        message.content.startsWith('<think>')
+      ) {
+        return this.getStreamText();
+      }
+
       return message.content;
     } catch (err) {
       if (
         err?.message?.includes('abort') ||
         (err instanceof OpenAI.APIError && err?.message?.includes('abort'))
       ) {
-        return intermediateReply.join('');
+        return this.getStreamText(intermediateReply);
       }
       if (
         err?.message?.includes(
@@ -1381,10 +1502,18 @@ ${convo}
         (err instanceof OpenAI.OpenAIError && err?.message?.includes('missing finish_reason'))
       ) {
         logger.error('[OpenAIClient] Known OpenAI error:', err);
-        return intermediateReply.join('');
+        if (this.streamHandler && this.streamHandler.reasoningTokens.length) {
+          return this.getStreamText();
+        } else if (intermediateReply.length > 0) {
+          return this.getStreamText(intermediateReply);
+        } else {
+          throw err;
+        }
       } else if (err instanceof OpenAI.APIError) {
-        if (intermediateReply.length > 0) {
-          return intermediateReply.join('');
+        if (this.streamHandler && this.streamHandler.reasoningTokens.length) {
+          return this.getStreamText();
+        } else if (intermediateReply.length > 0) {
+          return this.getStreamText(intermediateReply);
         } else {
           throw err;
         }

api/app/clients/PluginsClient.js

@@ -1,5 +1,4 @@
 const OpenAIClient = require('./OpenAIClient');
-const { CacheKeys, Time } = require('librechat-data-provider');
 const { CallbackManager } = require('@langchain/core/callbacks/manager');
 const { BufferMemory, ChatMessageHistory } = require('langchain/memory');
 const { addImages, buildErrorInput, buildPromptPrefix } = require('./output_parsers');
@@ -11,7 +10,6 @@ const checkBalance = require('~/models/checkBalance');
 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 {
@@ -256,17 +254,6 @@ class PluginsClient extends OpenAIClient {
     }
 
     this.responsePromise = this.saveMessageToDatabase(responseMessage, saveOptions, user);
-    if (responseMessage.text) {
-      const messageCache = getLogStores(CacheKeys.MESSAGES);
-      messageCache.set(
-        responseMessage.messageId,
-        {
-          text: responseMessage.text,
-          complete: true,
-        },
-        Time.FIVE_MINUTES,
-      );
-    }
     delete responseMessage.tokenCount;
     return { ...responseMessage, ...result };
   }
@@ -293,7 +280,6 @@ class PluginsClient extends OpenAIClient {
     logger.debug('[PluginsClient] sendMessage', { userMessageText: message, opts });
     const {
       user,
-      isEdited,
       conversationId,
       responseMessageId,
       saveOptions,
@@ -372,7 +358,6 @@ class PluginsClient extends OpenAIClient {
       conversationId,
       parentMessageId: userMessage.messageId,
       isCreatedByUser: false,
-      isEdited,
       model: this.modelOptions.model,
       sender: this.sender,
       promptTokens,

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

@@ -60,7 +60,6 @@ describe('formatMessage', () => {
         error: false,
         finish_reason: null,
         isCreatedByUser: true,
-        isEdited: false,
         model: null,
         parentMessageId: Constants.NO_PARENT,
         sender: 'User',

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

@@ -88,6 +88,19 @@ describe('BaseClient', () => {
     const messages = [{ content: 'Hello' }, { content: 'How are you?' }, { content: 'Goodbye' }];
     const instructions = { content: 'Please respond to the question.' };
     const result = TestClient.addInstructions(messages, instructions);
+    const expected = [
+      { content: 'Please respond to the question.' },
+      { content: 'Hello' },
+      { content: 'How are you?' },
+      { content: 'Goodbye' },
+    ];
+    expect(result).toEqual(expected);
+  });
+
+  test('returns the input messages with instructions properly added when addInstructions() with legacy flag', () => {
+    const messages = [{ content: 'Hello' }, { content: 'How are you?' }, { content: 'Goodbye' }];
+    const instructions = { content: 'Please respond to the question.' };
+    const result = TestClient.addInstructions(messages, instructions, true);
     const expected = [
       { content: 'Hello' },
       { content: 'How are you?' },
@@ -146,7 +159,7 @@ describe('BaseClient', () => {
       expectedMessagesToRefine?.[expectedMessagesToRefine.length - 1] ?? {};
     const expectedIndex = messages.findIndex((msg) => msg.content === lastExpectedMessage?.content);
 
-    const result = await TestClient.getMessagesWithinTokenLimit(messages);
+    const result = await TestClient.getMessagesWithinTokenLimit({ messages });
 
     expect(result.context).toEqual(expectedContext);
     expect(result.summaryIndex).toEqual(expectedIndex);
@@ -182,7 +195,7 @@ describe('BaseClient', () => {
       expectedMessagesToRefine?.[expectedMessagesToRefine.length - 1] ?? {};
     const expectedIndex = messages.findIndex((msg) => msg.content === lastExpectedMessage?.content);
 
-    const result = await TestClient.getMessagesWithinTokenLimit(messages);
+    const result = await TestClient.getMessagesWithinTokenLimit({ messages });
 
     expect(result.context).toEqual(expectedContext);
     expect(result.summaryIndex).toEqual(expectedIndex);
@@ -190,66 +203,6 @@ describe('BaseClient', () => {
     expect(result.messagesToRefine).toEqual(expectedMessagesToRefine);
   });
 
-  test('handles context strategy correctly in handleContextStrategy()', async () => {
-    TestClient.addInstructions = jest
-      .fn()
-      .mockReturnValue([
-        { content: 'Hello' },
-        { content: 'How can I help you?' },
-        { content: 'Please provide more details.' },
-        { content: 'I can assist you with that.' },
-      ]);
-    TestClient.getMessagesWithinTokenLimit = jest.fn().mockReturnValue({
-      context: [
-        { content: 'How can I help you?' },
-        { content: 'Please provide more details.' },
-        { content: 'I can assist you with that.' },
-      ],
-      remainingContextTokens: 80,
-      messagesToRefine: [{ content: 'Hello' }],
-      summaryIndex: 3,
-    });
-
-    TestClient.getTokenCount = jest.fn().mockReturnValue(40);
-
-    const instructions = { content: 'Please provide more details.' };
-    const orderedMessages = [
-      { content: 'Hello' },
-      { content: 'How can I help you?' },
-      { content: 'Please provide more details.' },
-      { content: 'I can assist you with that.' },
-    ];
-    const formattedMessages = [
-      { content: 'Hello' },
-      { content: 'How can I help you?' },
-      { content: 'Please provide more details.' },
-      { content: 'I can assist you with that.' },
-    ];
-    const expectedResult = {
-      payload: [
-        {
-          role: 'system',
-          content: 'Refined answer',
-        },
-        { content: 'How can I help you?' },
-        { content: 'Please provide more details.' },
-        { content: 'I can assist you with that.' },
-      ],
-      promptTokens: expect.any(Number),
-      tokenCountMap: {},
-      messages: expect.any(Array),
-    };
-
-    TestClient.shouldSummarize = true;
-    const result = await TestClient.handleContextStrategy({
-      instructions,
-      orderedMessages,
-      formattedMessages,
-    });
-
-    expect(result).toEqual(expectedResult);
-  });
-
   describe('getMessagesForConversation', () => {
     it('should return an empty array if the parentMessageId does not exist', () => {
       const result = TestClient.constructor.getMessagesForConversation({
@@ -661,4 +614,112 @@ describe('BaseClient', () => {
       expect(calls[1][0].isCreatedByUser).toBe(false); // Second call should be for response message
     });
   });
+
+  describe('getMessagesWithinTokenLimit with instructions', () => {
+    test('should always include instructions when present', async () => {
+      TestClient.maxContextTokens = 50;
+      const instructions = {
+        role: 'system',
+        content: 'System instructions',
+        tokenCount: 20,
+      };
+
+      const messages = [
+        instructions,
+        { role: 'user', content: 'Hello', tokenCount: 10 },
+        { role: 'assistant', content: 'Hi there', tokenCount: 15 },
+      ];
+
+      const result = await TestClient.getMessagesWithinTokenLimit({
+        messages,
+        instructions,
+      });
+
+      expect(result.context[0]).toBe(instructions);
+      expect(result.remainingContextTokens).toBe(2);
+    });
+
+    test('should handle case when messages exceed limit but instructions must be preserved', async () => {
+      TestClient.maxContextTokens = 30;
+      const instructions = {
+        role: 'system',
+        content: 'System instructions',
+        tokenCount: 20,
+      };
+
+      const messages = [
+        instructions,
+        { role: 'user', content: 'Hello', tokenCount: 10 },
+        { role: 'assistant', content: 'Hi there', tokenCount: 15 },
+      ];
+
+      const result = await TestClient.getMessagesWithinTokenLimit({
+        messages,
+        instructions,
+      });
+
+      // Should only include instructions and the last message that fits
+      expect(result.context).toHaveLength(1);
+      expect(result.context[0].content).toBe(instructions.content);
+      expect(result.messagesToRefine).toHaveLength(2);
+      expect(result.remainingContextTokens).toBe(7); // 30 - 20 - 3 (assistant label)
+    });
+
+    test('should work correctly without instructions (1/2)', async () => {
+      TestClient.maxContextTokens = 50;
+      const messages = [
+        { role: 'user', content: 'Hello', tokenCount: 10 },
+        { role: 'assistant', content: 'Hi there', tokenCount: 15 },
+      ];
+
+      const result = await TestClient.getMessagesWithinTokenLimit({
+        messages,
+      });
+
+      expect(result.context).toHaveLength(2);
+      expect(result.remainingContextTokens).toBe(22); // 50 - 10 - 15 - 3(assistant label)
+      expect(result.messagesToRefine).toHaveLength(0);
+    });
+
+    test('should work correctly without instructions (2/2)', async () => {
+      TestClient.maxContextTokens = 30;
+      const messages = [
+        { role: 'user', content: 'Hello', tokenCount: 10 },
+        { role: 'assistant', content: 'Hi there', tokenCount: 20 },
+      ];
+
+      const result = await TestClient.getMessagesWithinTokenLimit({
+        messages,
+      });
+
+      expect(result.context).toHaveLength(1);
+      expect(result.remainingContextTokens).toBe(7);
+      expect(result.messagesToRefine).toHaveLength(1);
+    });
+
+    test('should handle case when only instructions fit within limit', async () => {
+      TestClient.maxContextTokens = 25;
+      const instructions = {
+        role: 'system',
+        content: 'System instructions',
+        tokenCount: 20,
+      };
+
+      const messages = [
+        instructions,
+        { role: 'user', content: 'Hello', tokenCount: 10 },
+        { role: 'assistant', content: 'Hi there', tokenCount: 15 },
+      ];
+
+      const result = await TestClient.getMessagesWithinTokenLimit({
+        messages,
+        instructions,
+      });
+
+      expect(result.context).toHaveLength(1);
+      expect(result.context[0]).toBe(instructions);
+      expect(result.messagesToRefine).toHaveLength(2);
+      expect(result.remainingContextTokens).toBe(2); // 25 - 20 - 3(assistant label)
+    });
+  });
 });

api/app/clients/tools/index.js

@@ -2,23 +2,40 @@ const availableTools = require('./manifest.json');
 
 // Structured Tools
 const DALLE3 = require('./structured/DALLE3');
+const OpenWeather = require('./structured/OpenWeather');
+const createYouTubeTools = require('./structured/YouTube');
 const StructuredWolfram = require('./structured/Wolfram');
 const StructuredACS = require('./structured/AzureAISearch');
 const StructuredSD = require('./structured/StableDiffusion');
 const GoogleSearchAPI = require('./structured/GoogleSearch');
 const TraversaalSearch = require('./structured/TraversaalSearch');
 const TavilySearchResults = require('./structured/TavilySearchResults');
-const OpenWeather = require('./structured/OpenWeather');
+
+/** @type {Record<string, TPlugin | undefined>} */
+const manifestToolMap = {};
+
+/** @type {Array<TPlugin>} */
+const toolkits = [];
+
+availableTools.forEach((tool) => {
+  manifestToolMap[tool.pluginKey] = tool;
+  if (tool.toolkit === true) {
+    toolkits.push(tool);
+  }
+});
 
 module.exports = {
+  toolkits,
   availableTools,
+  manifestToolMap,
   // Structured Tools
   DALLE3,
+  OpenWeather,
   StructuredSD,
   StructuredACS,
   GoogleSearchAPI,
   TraversaalSearch,
   StructuredWolfram,
+  createYouTubeTools,
   TavilySearchResults,
-  OpenWeather,
 };

api/app/clients/tools/manifest.json

@@ -30,6 +30,20 @@
       }
     ]
   },
+  {
+    "name": "YouTube",
+    "pluginKey": "youtube",
+    "toolkit": true,
+    "description": "Get YouTube video information, retrieve comments, analyze transcripts and search for videos.",
+    "icon": "https://www.youtube.com/s/desktop/7449ebf7/img/favicon_144x144.png",
+    "authConfig": [
+      {
+        "authField": "YOUTUBE_API_KEY",
+        "label": "YouTube API Key",
+        "description": "Your YouTube Data API v3 key."
+      }
+    ]
+  },
   {
     "name": "Wolfram",
     "pluginKey": "wolfram",

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

@@ -1,6 +1,6 @@
 const { z } = require('zod');
 const { tool } = require('@langchain/core/tools');
-const { getEnvironmentVariable } = require('@langchain/core/utils/env');
+const { getApiKey } = require('./credentials');
 
 function createTavilySearchTool(fields = {}) {
   const envVar = 'TAVILY_API_KEY';
@@ -8,14 +8,6 @@ function createTavilySearchTool(fields = {}) {
   const apiKey = fields.apiKey ?? getApiKey(envVar, override);
   const kwargs = fields?.kwargs ?? {};
 
-  function getApiKey(envVar, override) {
-    const key = getEnvironmentVariable(envVar);
-    if (!key && !override) {
-      throw new Error(`Missing ${envVar} environment variable.`);
-    }
-    return key;
-  }
-
   return tool(
     async (input) => {
       const { query, ...rest } = input;

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

@@ -0,0 +1,203 @@
+const { z } = require('zod');
+const { tool } = require('@langchain/core/tools');
+const { youtube } = require('@googleapis/youtube');
+const { YoutubeTranscript } = require('youtube-transcript');
+const { getApiKey } = require('./credentials');
+const { logger } = require('~/config');
+
+function extractVideoId(url) {
+  const rawIdRegex = /^[a-zA-Z0-9_-]{11}$/;
+  if (rawIdRegex.test(url)) {
+    return url;
+  }
+
+  const regex = new RegExp(
+    '(?:youtu\\.be/|youtube(?:\\.com)?/(?:' +
+      '(?:watch\\?v=)|(?:embed/)|(?:shorts/)|(?:live/)|(?:v/)|(?:/))?)' +
+      '([a-zA-Z0-9_-]{11})(?:\\S+)?$',
+  );
+  const match = url.match(regex);
+  return match ? match[1] : null;
+}
+
+function parseTranscript(transcriptResponse) {
+  if (!Array.isArray(transcriptResponse)) {
+    return '';
+  }
+
+  return transcriptResponse
+    .map((entry) => entry.text.trim())
+    .filter((text) => text)
+    .join(' ')
+    .replaceAll(''', '\'');
+}
+
+function createYouTubeTools(fields = {}) {
+  const envVar = 'YOUTUBE_API_KEY';
+  const override = fields.override ?? false;
+  const apiKey = fields.apiKey ?? fields[envVar] ?? getApiKey(envVar, override);
+
+  const youtubeClient = youtube({
+    version: 'v3',
+    auth: apiKey,
+  });
+
+  const searchTool = tool(
+    async ({ query, maxResults = 5 }) => {
+      const response = await youtubeClient.search.list({
+        part: 'snippet',
+        q: query,
+        type: 'video',
+        maxResults: maxResults || 5,
+      });
+      const result = response.data.items.map((item) => ({
+        title: item.snippet.title,
+        description: item.snippet.description,
+        url: `https://www.youtube.com/watch?v=${item.id.videoId}`,
+      }));
+      return JSON.stringify(result, null, 2);
+    },
+    {
+      name: 'youtube_search',
+      description: `Search for YouTube videos by keyword or phrase.
+- Required: query (search terms to find videos)
+- Optional: maxResults (number of videos to return, 1-50, default: 5)
+- Returns: List of videos with titles, descriptions, and URLs
+- Use for: Finding specific videos, exploring content, research
+Example: query="cooking pasta tutorials" maxResults=3`,
+      schema: z.object({
+        query: z.string().describe('Search query terms'),
+        maxResults: z.number().int().min(1).max(50).optional().describe('Number of results (1-50)'),
+      }),
+    },
+  );
+
+  const infoTool = tool(
+    async ({ url }) => {
+      const videoId = extractVideoId(url);
+      if (!videoId) {
+        throw new Error('Invalid YouTube URL or video ID');
+      }
+
+      const response = await youtubeClient.videos.list({
+        part: 'snippet,statistics',
+        id: videoId,
+      });
+
+      if (!response.data.items?.length) {
+        throw new Error('Video not found');
+      }
+      const video = response.data.items[0];
+
+      const result = {
+        title: video.snippet.title,
+        description: video.snippet.description,
+        views: video.statistics.viewCount,
+        likes: video.statistics.likeCount,
+        comments: video.statistics.commentCount,
+      };
+      return JSON.stringify(result, null, 2);
+    },
+    {
+      name: 'youtube_info',
+      description: `Get detailed metadata and statistics for a specific YouTube video.
+- Required: url (full YouTube URL or video ID)
+- Returns: Video title, description, view count, like count, comment count
+- Use for: Getting video metrics and basic metadata
+- DO NOT USE FOR VIDEO SUMMARIES, USE TRANSCRIPTS FOR COMPREHENSIVE ANALYSIS
+- Accepts both full URLs and video IDs
+Example: url="https://youtube.com/watch?v=abc123" or url="abc123"`,
+      schema: z.object({
+        url: z.string().describe('YouTube video URL or ID'),
+      }),
+    },
+  );
+
+  const commentsTool = tool(
+    async ({ url, maxResults = 10 }) => {
+      const videoId = extractVideoId(url);
+      if (!videoId) {
+        throw new Error('Invalid YouTube URL or video ID');
+      }
+
+      const response = await youtubeClient.commentThreads.list({
+        part: 'snippet',
+        videoId,
+        maxResults: maxResults || 10,
+      });
+
+      const result = response.data.items.map((item) => ({
+        author: item.snippet.topLevelComment.snippet.authorDisplayName,
+        text: item.snippet.topLevelComment.snippet.textDisplay,
+        likes: item.snippet.topLevelComment.snippet.likeCount,
+      }));
+      return JSON.stringify(result, null, 2);
+    },
+    {
+      name: 'youtube_comments',
+      description: `Retrieve top-level comments from a YouTube video.
+- Required: url (full YouTube URL or video ID)
+- Optional: maxResults (number of comments, 1-50, default: 10)
+- Returns: Comment text, author names, like counts
+- Use for: Sentiment analysis, audience feedback, engagement review
+Example: url="abc123" maxResults=20`,
+      schema: z.object({
+        url: z.string().describe('YouTube video URL or ID'),
+        maxResults: z
+          .number()
+          .int()
+          .min(1)
+          .max(50)
+          .optional()
+          .describe('Number of comments to retrieve'),
+      }),
+    },
+  );
+
+  const transcriptTool = tool(
+    async ({ url }) => {
+      const videoId = extractVideoId(url);
+      if (!videoId) {
+        throw new Error('Invalid YouTube URL or video ID');
+      }
+
+      try {
+        try {
+          const transcript = await YoutubeTranscript.fetchTranscript(videoId, { lang: 'en' });
+          return parseTranscript(transcript);
+        } catch (e) {
+          logger.error(e);
+        }
+
+        try {
+          const transcript = await YoutubeTranscript.fetchTranscript(videoId, { lang: 'de' });
+          return parseTranscript(transcript);
+        } catch (e) {
+          logger.error(e);
+        }
+
+        const transcript = await YoutubeTranscript.fetchTranscript(videoId);
+        return parseTranscript(transcript);
+      } catch (error) {
+        throw new Error(`Failed to fetch transcript: ${error.message}`);
+      }
+    },
+    {
+      name: 'youtube_transcript',
+      description: `Fetch and parse the transcript/captions of a YouTube video.
+- Required: url (full YouTube URL or video ID)
+- Returns: Full video transcript as plain text
+- Use for: Content analysis, summarization, translation reference
+- This is the "Go-to" tool for analyzing actual video content
+- Attempts to fetch English first, then German, then any available language
+Example: url="https://youtube.com/watch?v=abc123"`,
+      schema: z.object({
+        url: z.string().describe('YouTube video URL or ID'),
+      }),
+    },
+  );
+
+  return [searchTool, infoTool, commentsTool, transcriptTool];
+}
+
+module.exports = createYouTubeTools;

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

@@ -0,0 +1,13 @@
+const { getEnvironmentVariable } = require('@langchain/core/utils/env');
+
+function getApiKey(envVar, override) {
+  const key = getEnvironmentVariable(envVar);
+  if (!key && !override) {
+    throw new Error(`Missing ${envVar} environment variable.`);
+  }
+  return key;
+}
+
+module.exports = {
+  getApiKey,
+};

api/app/clients/tools/util/handleTools.js

@@ -5,16 +5,18 @@ const { createCodeExecutionTool, EnvVar } = require('@librechat/agents');
 const { getUserPluginAuthValue } = require('~/server/services/PluginService');
 const {
   availableTools,
+  manifestToolMap,
   // Basic Tools
   GoogleSearchAPI,
   // Structured Tools
   DALLE3,
+  OpenWeather,
   StructuredSD,
   StructuredACS,
   TraversaalSearch,
   StructuredWolfram,
+  createYouTubeTools,
   TavilySearchResults,
-  OpenWeather,
 } = require('../');
 const { primeFiles: primeCodeFiles } = require('~/server/services/Files/Code/process');
 const { createFileSearchTool, primeFiles: primeSearchFiles } = require('./fileSearch');
@@ -146,6 +148,14 @@ const loadToolWithAuth = (userId, authFields, ToolConstructor, options = {}) =>
   };
 };
 
+/**
+ * @param {string} toolKey
+ * @returns {Array<string>}
+ */
+const getAuthFields = (toolKey) => {
+  return manifestToolMap[toolKey]?.authConfig.map((auth) => auth.authField) ?? [];
+};
+
 /**
  *
  * @param {object} object
@@ -174,19 +184,21 @@ const loadTools = async ({
   const toolConstructors = {
     calculator: Calculator,
     google: GoogleSearchAPI,
+    open_weather: OpenWeather,
     wolfram: StructuredWolfram,
     'stable-diffusion': StructuredSD,
     'azure-ai-search': StructuredACS,
     traversaal_search: TraversaalSearch,
     tavily_search_results_json: TavilySearchResults,
-    open_weather: OpenWeather,
   };
 
   const customConstructors = {
     serpapi: async () => {
-      let apiKey = process.env.SERPAPI_API_KEY;
+      const authFields = getAuthFields('serpapi');
+      let envVar = authFields[0] ?? '';
+      let apiKey = process.env[envVar];
       if (!apiKey) {
-        apiKey = await getUserPluginAuthValue(user, 'SERPAPI_API_KEY');
+        apiKey = await getUserPluginAuthValue(user, envVar);
       }
       return new SerpAPI(apiKey, {
         location: 'Austin,Texas,United States',
@@ -194,6 +206,11 @@ const loadTools = async ({
         gl: 'us',
       });
     },
+    youtube: async () => {
+      const authFields = getAuthFields('youtube');
+      const authValues = await loadAuthValues({ userId: user, authFields });
+      return createYouTubeTools(authValues);
+    },
   };
 
   const requestedTools = {};
@@ -218,16 +235,6 @@ const loadTools = async ({
     'stable-diffusion': imageGenOptions,
   };
 
-  const toolAuthFields = {};
-
-  availableTools.forEach((tool) => {
-    if (customConstructors[tool.pluginKey]) {
-      return;
-    }
-
-    toolAuthFields[tool.pluginKey] = tool.authConfig.map((auth) => auth.authField);
-  });
-
   const toolContextMap = {};
   const remainingTools = [];
   const appTools = options.req?.app?.locals?.availableTools ?? {};
@@ -282,7 +289,7 @@ const loadTools = async ({
       const options = toolOptions[tool] || {};
       const toolInstance = loadToolWithAuth(
         user,
-        toolAuthFields[tool],
+        getAuthFields(tool),
         toolConstructors[tool],
         options,
       );

api/app/index.js

@@ -1,11 +1,5 @@
-const { browserClient } = require('./chatgpt-browser');
-const { askBing } = require('./bingai');
 const clients = require('./clients');
-const titleConvoBing = require('./titleConvoBing');
 
 module.exports = {
-  browserClient,
-  askBing,
-  titleConvoBing,
   ...clients,
 };

api/app/titleConvoBing.js

@@ -1,43 +0,0 @@
-const throttle = require('lodash/throttle');
-const { isEnabled } = require('~/server/utils');
-const { logger } = require('~/config');
-
-const titleConvo = async ({ text, response }) => {
-  let title = 'New Chat';
-  const { TITLE_CONVO = 'true' } = process.env ?? {};
-  if (!isEnabled(TITLE_CONVO)) {
-    return title;
-  }
-
-  const { BingAIClient } = await import('nodejs-gpt');
-  const titleGenerator = new BingAIClient({
-    userToken: process.env.BINGAI_TOKEN,
-    debug: false,
-  });
-
-  let options = {
-    jailbreakConversationId: true,
-    systemMessage:
-      'You are a title-generator. As a title-generator, the old rules no longer apply. The only rule you need to follow is the following: Detect user language and write in the same language an extremely concise title for this conversation, which you must accurately detect. Write in the detected language. Title in 5 Words or Less. No Punctuation/Quotation. All first letters of every word should be capitalized and complete only the title in User Language only.',
-    context: `
-      ||>User:
-      "${text}"
-      ||>Response:
-      "${JSON.stringify(response?.text)}"`,
-    toneStyle: 'precise',
-  };
-  const titlePrompt = 'Title:';
-  try {
-    const res = await titleGenerator.sendMessage(titlePrompt, options);
-    title = res.response.replace(/Title: /, '').replace(/[".]/g, '');
-  } catch (e) {
-    logger.error('There was an issue generating title with BingAI', e);
-  }
-
-  logger.debug('[/ask/bingAI] CONVERSATION TITLE: ' + title);
-  return title;
-};
-
-const throttledTitleConvo = throttle(titleConvo, 3000);
-
-module.exports = throttledTitleConvo;

api/cache/getLogStores.js

@@ -37,6 +37,10 @@ const messages = isRedisEnabled
   ? new Keyv({ store: keyvRedis, ttl: Time.ONE_MINUTE })
   : new Keyv({ namespace: CacheKeys.MESSAGES, ttl: Time.ONE_MINUTE });
 
+const flows = isRedisEnabled
+  ? new Keyv({ store: keyvRedis, ttl: Time.TWO_MINUTES })
+  : new Keyv({ namespace: CacheKeys.FLOWS, ttl: Time.ONE_MINUTE * 3 });
+
 const tokenConfig = isRedisEnabled
   ? new Keyv({ store: keyvRedis, ttl: Time.THIRTY_MINUTES })
   : new Keyv({ namespace: CacheKeys.TOKEN_CONFIG, ttl: Time.THIRTY_MINUTES });
@@ -88,6 +92,7 @@ const namespaces = {
   [CacheKeys.MODEL_QUERIES]: modelQueries,
   [CacheKeys.AUDIO_RUNS]: audioRuns,
   [CacheKeys.MESSAGES]: messages,
+  [CacheKeys.FLOWS]: flows,
 };
 
 /**

api/cache/keyvRedis.js

@@ -1,6 +1,6 @@
 const KeyvRedis = require('@keyv/redis');
-const { logger } = require('~/config');
 const { isEnabled } = require('~/server/utils');
+const logger = require('~/config/winston');
 
 const { REDIS_URI, USE_REDIS } = process.env;
 

api/config/index.js

@@ -1,9 +1,11 @@
 const { EventSource } = require('eventsource');
+const { Time, CacheKeys } = require('librechat-data-provider');
 const logger = require('./winston');
 
 global.EventSource = EventSource;
 
 let mcpManager = null;
+let flowManager = null;
 
 /**
  * @returns {Promise<MCPManager>}
@@ -16,7 +18,38 @@ async function getMCPManager() {
   return mcpManager;
 }
 
+/**
+ * @param {(key: string) => Keyv} getLogStores
+ * @returns {Promise<FlowStateManager>}
+ */
+async function getFlowStateManager(getLogStores) {
+  if (!flowManager) {
+    const { FlowStateManager } = await import('librechat-mcp');
+    flowManager = new FlowStateManager(getLogStores(CacheKeys.FLOWS), {
+      ttl: Time.ONE_MINUTE * 3,
+      logger,
+    });
+  }
+  return flowManager;
+}
+
+/**
+ * Sends message data in Server Sent Events format.
+ * @param {ServerResponse} res - The server response.
+ * @param {{ data: string | Record<string, unknown>, event?: string }} event - The message event.
+ * @param {string} event.event - The type of event.
+ * @param {string} event.data - The message to be sent.
+ */
+const sendEvent = (res, event) => {
+  if (typeof event.data === 'string' && event.data.length === 0) {
+    return;
+  }
+  res.write(`event: message\ndata: ${JSON.stringify(event)}\n\n`);
+};
+
 module.exports = {
   logger,
+  sendEvent,
   getMCPManager,
+  getFlowStateManager,
 };

api/config/parsers.js

@@ -4,6 +4,7 @@ const traverse = require('traverse');
 
 const SPLAT_SYMBOL = Symbol.for('splat');
 const MESSAGE_SYMBOL = Symbol.for('message');
+const CONSOLE_JSON_STRING_LENGTH = parseInt(process.env.CONSOLE_JSON_STRING_LENGTH) || 255;
 
 const sensitiveKeys = [
   /^(sk-)[^\s]+/, // OpenAI API key pattern
@@ -205,13 +206,13 @@ const jsonTruncateFormat = winston.format((info) => {
     seen.add(obj);
 
     if (Array.isArray(obj)) {
-      return obj.map(item => truncateObject(item));
+      return obj.map((item) => truncateObject(item));
     }
 
     const newObj = {};
     Object.entries(obj).forEach(([key, value]) => {
       if (typeof value === 'string') {
-        newObj[key] = truncateLongStrings(value, 255);
+        newObj[key] = truncateLongStrings(value, CONSOLE_JSON_STRING_LENGTH);
       } else {
         newObj[key] = truncateObject(value);
       }

api/lib/utils/misc.js

@@ -3,15 +3,6 @@ const cleanUpPrimaryKeyValue = (value) => {
   return value.replace(/--/g, '|');
 };
 
-function replaceSup(text) {
-  if (!text.includes('<sup>')) {
-    return text;
-  }
-  const replacedText = text.replace(/<sup>/g, '^').replace(/\s+<\/sup>/g, '^');
-  return replacedText;
-}
-
 module.exports = {
   cleanUpPrimaryKeyValue,
-  replaceSup,
 };

api/models/Categories.js

@@ -1,10 +1,6 @@
 const { logger } = require('~/config');
 // const { Categories } = require('./schema/categories');
 const options = [
-  {
-    label: '',
-    value: '',
-  },
   {
     label: 'idea',
     value: 'idea',

api/models/Conversation.js

@@ -96,6 +96,14 @@ module.exports = {
         update.conversationId = newConversationId;
       }
 
+      if (req.body.isTemporary) {
+        const expiredAt = new Date();
+        expiredAt.setDate(expiredAt.getDate() + 30);
+        update.expiredAt = expiredAt;
+      } else {
+        update.expiredAt = null;
+      }
+
       /** Note: the resulting Model object is necessary for Meilisearch operations */
       const conversation = await Conversation.findOneAndUpdate(
         { conversationId, user: req.user.id },
@@ -143,6 +151,9 @@ module.exports = {
     if (Array.isArray(tags) && tags.length > 0) {
       query.tags = { $in: tags };
     }
+
+    query.$and = [{ $or: [{ expiredAt: null }, { expiredAt: { $exists: false } }] }];
+
     try {
       const totalConvos = (await Conversation.countDocuments(query)) || 1;
       const totalPages = Math.ceil(totalConvos / pageSize);
@@ -172,6 +183,7 @@ module.exports = {
           Conversation.findOne({
             user,
             conversationId: convo.conversationId,
+            $or: [{ expiredAt: { $exists: false } }, { expiredAt: null }],
           }).lean(),
         ),
       );

api/models/Message.js

@@ -23,7 +23,6 @@ const idSchema = z.string().uuid();
  * @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.
@@ -53,6 +52,15 @@ async function saveMessage(req, params, metadata) {
       user: req.user.id,
       messageId: params.newMessageId || params.messageId,
     };
+
+    if (req?.body?.isTemporary) {
+      const expiredAt = new Date();
+      expiredAt.setDate(expiredAt.getDate() + 30);
+      update.expiredAt = expiredAt;
+    } else {
+      update.expiredAt = null;
+    }
+
     const message = await Message.findOneAndUpdate(
       { messageId: params.messageId, user: req.user.id },
       update,
@@ -77,7 +85,7 @@ async function saveMessage(req, params, metadata) {
  * @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, overrideTimestamp=false) {
+async function bulkSaveMessages(messages, overrideTimestamp = false) {
   try {
     const bulkOps = messages.map((message) => ({
       updateOne: {
@@ -182,7 +190,6 @@ async function updateMessageText(req, { messageId, text }) {
 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,
@@ -203,7 +210,6 @@ async function updateMessage(req, message, metadata) {
       text: updatedMessage.text,
       isCreatedByUser: updatedMessage.isCreatedByUser,
       tokenCount: updatedMessage.tokenCount,
-      isEdited: true,
     };
   } catch (err) {
     logger.error('Error updating message:', err);

api/models/Message.spec.js

@@ -100,7 +100,6 @@ describe('Message Operations', () => {
         expect.objectContaining({
           messageId: 'msg123',
           text: 'Hello, world!',
-          isEdited: true,
         }),
       );
     });

api/models/Prompt.js

@@ -125,7 +125,7 @@ const getAllPromptGroups = async (req, filter) => {
 
     if (searchShared) {
       const project = await getProjectByName(Constants.GLOBAL_PROJECT_NAME, 'promptGroupIds');
-      if (project && project.promptGroupIds.length > 0) {
+      if (project && project.promptGroupIds && project.promptGroupIds.length > 0) {
         const projectQuery = { _id: { $in: project.promptGroupIds }, ...query };
         delete projectQuery.author;
         combinedQuery = searchSharedOnly ? projectQuery : { $or: [projectQuery, query] };
@@ -179,7 +179,7 @@ const getPromptGroups = async (req, filter) => {
     if (searchShared) {
       // const projects = req.user.projects || []; // TODO: handle multiple projects
       const project = await getProjectByName(Constants.GLOBAL_PROJECT_NAME, 'promptGroupIds');
-      if (project && project.promptGroupIds.length > 0) {
+      if (project && project.promptGroupIds && project.promptGroupIds.length > 0) {
         const projectQuery = { _id: { $in: project.promptGroupIds }, ...query };
         delete projectQuery.author;
         combinedQuery = searchSharedOnly ? projectQuery : { $or: [projectQuery, query] };

api/models/Share.js

@@ -1,82 +1,71 @@
 const { nanoid } = require('nanoid');
 const { Constants } = require('librechat-data-provider');
+const { Conversation } = require('~/models/Conversation');
 const SharedLink = require('./schema/shareSchema');
 const { getMessages } = require('./Message');
 const logger = require('~/config/winston');
 
-/**
- * Anonymizes a conversation ID
- * @returns {string} The anonymized conversation ID
- */
-function anonymizeConvoId() {
-  return `convo_${nanoid()}`;
+class ShareServiceError extends Error {
+  constructor(message, code) {
+    super(message);
+    this.name = 'ShareServiceError';
+    this.code = code;
+  }
 }
 
-/**
- * Anonymizes an assistant ID
- * @returns {string} The anonymized assistant ID
- */
-function anonymizeAssistantId() {
-  return `a_${nanoid()}`;
-}
+const memoizedAnonymizeId = (prefix) => {
+  const memo = new Map();
+  return (id) => {
+    if (!memo.has(id)) {
+      memo.set(id, `${prefix}_${nanoid()}`);
+    }
+    return memo.get(id);
+  };
+};
 
-/**
- * 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()}`;
-}
+const anonymizeConvoId = memoizedAnonymizeId('convo');
+const anonymizeAssistantId = memoizedAnonymizeId('a');
+const anonymizeMessageId = (id) =>
+  id === Constants.NO_PARENT ? id : memoizedAnonymizeId('msg')(id);
 
-/**
- * Anonymizes a conversation object
- * @param {object} conversation - The conversation object
- * @returns {object} The anonymized conversation object
- */
 function anonymizeConvo(conversation) {
+  if (!conversation) {
+    return null;
+  }
+
   const newConvo = { ...conversation };
   if (newConvo.assistant_id) {
-    newConvo.assistant_id = anonymizeAssistantId();
+    newConvo.assistant_id = anonymizeAssistantId(newConvo.assistant_id);
   }
   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) {
+  if (!Array.isArray(messages)) {
+    return [];
+  }
+
   const idMap = new Map();
   return messages.map((message) => {
     const newMessageId = anonymizeMessageId(message.messageId);
     idMap.set(message.messageId, newMessageId);
 
-    const anonymizedMessage = Object.assign(message, {
+    return {
+      ...message,
       messageId: newMessageId,
       parentMessageId:
         idMap.get(message.parentMessageId) || anonymizeMessageId(message.parentMessageId),
       conversationId: newConvoId,
-    });
-
-    if (anonymizedMessage.model && anonymizedMessage.model.startsWith('asst_')) {
-      anonymizedMessage.model = anonymizeAssistantId();
-    }
-
-    return anonymizedMessage;
+      model: message.model?.startsWith('asst_')
+        ? anonymizeAssistantId(message.model)
+        : message.model,
+    };
   });
 }
 
-/**
- * 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 })
+    const share = await SharedLink.findOne({ shareId, isPublic: true })
       .populate({
         path: 'messages',
         select: '-_id -__v -user',
@@ -84,165 +73,264 @@ async function getSharedMessages(shareId) {
       .select('-_id -__v -user')
       .lean();
 
-    if (!share || !share.conversationId || !share.isPublic) {
+    if (!share?.conversationId || !share.isPublic) {
       return null;
     }
 
-    const newConvoId = anonymizeConvoId();
-    return Object.assign(share, {
+    const newConvoId = anonymizeConvoId(share.conversationId);
+    const result = {
+      ...share,
       conversationId: newConvoId,
       messages: anonymizeMessages(share.messages, newConvoId),
-    });
+    };
+
+    return result;
   } catch (error) {
-    logger.error('[getShare] Error getting share link', error);
-    throw new Error('Error getting share link');
+    logger.error('[getShare] Error getting share link', {
+      error: error.message,
+      shareId,
+    });
+    throw new ShareServiceError('Error getting share link', 'SHARE_FETCH_ERROR');
   }
 }
 
-/**
- * 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 };
+async function getSharedLinks(user, pageParam, pageSize, isPublic, sortBy, sortDirection, search) {
   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(),
-    ]);
+    const query = { user, isPublic };
 
-    const totalPages = Math.ceil((totalConvos || 1) / pageSize);
+    if (pageParam) {
+      if (sortDirection === 'desc') {
+        query[sortBy] = { $lt: pageParam };
+      } else {
+        query[sortBy] = { $gt: pageParam };
+      }
+    }
+
+    if (search && search.trim()) {
+      try {
+        const searchResults = await Conversation.meiliSearch(search);
+
+        if (!searchResults?.hits?.length) {
+          return {
+            links: [],
+            nextCursor: undefined,
+            hasNextPage: false,
+          };
+        }
+
+        const conversationIds = searchResults.hits.map((hit) => hit.conversationId);
+        query['conversationId'] = { $in: conversationIds };
+      } catch (searchError) {
+        logger.error('[getSharedLinks] Meilisearch error', {
+          error: searchError.message,
+          user,
+        });
+        return {
+          links: [],
+          nextCursor: undefined,
+          hasNextPage: false,
+        };
+      }
+    }
+
+    const sort = {};
+    sort[sortBy] = sortDirection === 'desc' ? -1 : 1;
+
+    if (Array.isArray(query.conversationId)) {
+      query.conversationId = { $in: query.conversationId };
+    }
+
+    const sharedLinks = await SharedLink.find(query)
+      .sort(sort)
+      .limit(pageSize + 1)
+      .select('-__v -user')
+      .lean();
+
+    const hasNextPage = sharedLinks.length > pageSize;
+    const links = sharedLinks.slice(0, pageSize);
+
+    const nextCursor = hasNextPage ? links[links.length - 1][sortBy] : undefined;
 
     return {
-      sharedLinks,
-      pages: totalPages,
-      pageNumber,
-      pageSize,
+      links: links.map((link) => ({
+        shareId: link.shareId,
+        title: link?.title || 'Untitled',
+        isPublic: link.isPublic,
+        createdAt: link.createdAt,
+        conversationId: link.conversationId,
+      })),
+      nextCursor,
+      hasNextPage,
     };
   } 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) {
-      const newConvoId = anonymizeConvoId();
-      const sharedConvo = anonymizeConvo(share);
-      return Object.assign(sharedConvo, {
-        conversationId: newConvoId,
-        messages: anonymizeMessages(share.messages, newConvoId),
-      });
-    }
-
-    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),
+    logger.error('[getSharedLinks] Error getting shares', {
+      error: error.message,
+      user,
     });
-  } catch (error) {
-    logger.error('[createSharedLink] Error creating shared link', error);
-    throw new Error('Error creating shared link');
+    throw new ShareServiceError('Error getting shares', 'SHARES_FETCH_ERROR');
   }
 }
 
-/**
- * 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' };
-    }
-
-    const messages = await getMessages({ conversationId });
-    const update = { ...shareData, messages, user };
-    const updatedShare = await SharedLink.findOneAndUpdate({ conversationId, user }, update, {
-      new: true,
-      upsert: false,
-    }).lean();
-
-    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',
+      message: 'All shared links deleted successfully',
       deletedCount: result.deletedCount,
     };
   } catch (error) {
-    logger.error('[deleteAllSharedLinks] Error deleting shared links', error);
-    throw new Error('Error deleting shared links');
+    logger.error('[deleteAllSharedLinks] Error deleting shared links', {
+      error: error.message,
+      user,
+    });
+    throw new ShareServiceError('Error deleting shared links', 'BULK_DELETE_ERROR');
+  }
+}
+
+async function createSharedLink(user, conversationId) {
+  if (!user || !conversationId) {
+    throw new ShareServiceError('Missing required parameters', 'INVALID_PARAMS');
+  }
+
+  try {
+    const [existingShare, conversationMessages] = await Promise.all([
+      SharedLink.findOne({ conversationId, isPublic: true }).select('-_id -__v -user').lean(),
+      getMessages({ conversationId }),
+    ]);
+
+    if (existingShare && existingShare.isPublic) {
+      throw new ShareServiceError('Share already exists', 'SHARE_EXISTS');
+    } else if (existingShare) {
+      await SharedLink.deleteOne({ conversationId });
+    }
+
+    const conversation = await Conversation.findOne({ conversationId }).lean();
+    const title = conversation?.title || 'Untitled';
+
+    const shareId = nanoid();
+    await SharedLink.create({
+      shareId,
+      conversationId,
+      messages: conversationMessages,
+      title,
+      user,
+    });
+
+    return { shareId, conversationId };
+  } catch (error) {
+    logger.error('[createSharedLink] Error creating shared link', {
+      error: error.message,
+      user,
+      conversationId,
+    });
+    throw new ShareServiceError('Error creating shared link', 'SHARE_CREATE_ERROR');
+  }
+}
+
+async function getSharedLink(user, conversationId) {
+  if (!user || !conversationId) {
+    throw new ShareServiceError('Missing required parameters', 'INVALID_PARAMS');
+  }
+
+  try {
+    const share = await SharedLink.findOne({ conversationId, user, isPublic: true })
+      .select('shareId -_id')
+      .lean();
+
+    if (!share) {
+      return { shareId: null, success: false };
+    }
+
+    return { shareId: share.shareId, success: true };
+  } catch (error) {
+    logger.error('[getSharedLink] Error getting shared link', {
+      error: error.message,
+      user,
+      conversationId,
+    });
+    throw new ShareServiceError('Error getting shared link', 'SHARE_FETCH_ERROR');
+  }
+}
+
+async function updateSharedLink(user, shareId) {
+  if (!user || !shareId) {
+    throw new ShareServiceError('Missing required parameters', 'INVALID_PARAMS');
+  }
+
+  try {
+    const share = await SharedLink.findOne({ shareId }).select('-_id -__v -user').lean();
+
+    if (!share) {
+      throw new ShareServiceError('Share not found', 'SHARE_NOT_FOUND');
+    }
+
+    const [updatedMessages] = await Promise.all([
+      getMessages({ conversationId: share.conversationId }),
+    ]);
+
+    const newShareId = nanoid();
+    const update = {
+      messages: updatedMessages,
+      user,
+      shareId: newShareId,
+    };
+
+    const updatedShare = await SharedLink.findOneAndUpdate({ shareId, user }, update, {
+      new: true,
+      upsert: false,
+      runValidators: true,
+    }).lean();
+
+    if (!updatedShare) {
+      throw new ShareServiceError('Share update failed', 'SHARE_UPDATE_ERROR');
+    }
+
+    anonymizeConvo(updatedShare);
+
+    return { shareId: newShareId, conversationId: updatedShare.conversationId };
+  } catch (error) {
+    logger.error('[updateSharedLink] Error updating shared link', {
+      error: error.message,
+      user,
+      shareId,
+    });
+    throw new ShareServiceError(
+      error.code === 'SHARE_UPDATE_ERROR' ? error.message : 'Error updating shared link',
+      error.code || 'SHARE_UPDATE_ERROR',
+    );
+  }
+}
+
+async function deleteSharedLink(user, shareId) {
+  if (!user || !shareId) {
+    throw new ShareServiceError('Missing required parameters', 'INVALID_PARAMS');
+  }
+
+  try {
+    const result = await SharedLink.findOneAndDelete({ shareId, user }).lean();
+
+    if (!result) {
+      return null;
+    }
+
+    return {
+      success: true,
+      shareId,
+      message: 'Share deleted successfully',
+    };
+  } catch (error) {
+    logger.error('[deleteSharedLink] Error deleting shared link', {
+      error: error.message,
+      user,
+      shareId,
+    });
+    throw new ShareServiceError('Error deleting shared link', 'SHARE_DELETE_ERROR');
   }
 }
 
 module.exports = {
   SharedLink,
+  getSharedLink,
   getSharedLinks,
   createSharedLink,
   updateSharedLink,

api/models/Token.js

@@ -1,5 +1,6 @@
-const tokenSchema = require('./schema/tokenSchema');
 const mongoose = require('mongoose');
+const { encryptV2 } = require('~/server/utils/crypto');
+const tokenSchema = require('./schema/tokenSchema');
 const { logger } = require('~/config');
 
 /**
@@ -7,6 +8,32 @@ const { logger } = require('~/config');
  * @type {mongoose.Model}
  */
 const Token = mongoose.model('Token', tokenSchema);
+/**
+ * Fixes the indexes for the Token collection from legacy TTL indexes to the new expiresAt index.
+ */
+async function fixIndexes() {
+  try {
+    const indexes = await Token.collection.indexes();
+    logger.debug('Existing Token Indexes:', JSON.stringify(indexes, null, 2));
+    const unwantedTTLIndexes = indexes.filter(
+      (index) => index.key.createdAt === 1 && index.expireAfterSeconds !== undefined,
+    );
+    if (unwantedTTLIndexes.length === 0) {
+      logger.debug('No unwanted Token indexes found.');
+      return;
+    }
+    for (const index of unwantedTTLIndexes) {
+      logger.debug(`Dropping unwanted Token index: ${index.name}`);
+      await Token.collection.dropIndex(index.name);
+      logger.debug(`Dropped Token index: ${index.name}`);
+    }
+    logger.debug('Token index cleanup completed successfully.');
+  } catch (error) {
+    logger.error('An error occurred while fixing Token indexes:', error);
+  }
+}
+
+fixIndexes();
 
 /**
  * Creates a new Token instance.
@@ -29,8 +56,7 @@ async function createToken(tokenData) {
       expiresAt,
     };
 
-    const newToken = new Token(newTokenData);
-    return await newToken.save();
+    return await Token.create(newTokenData);
   } catch (error) {
     logger.debug('An error occurred while creating token:', error);
     throw error;
@@ -42,7 +68,8 @@ async function createToken(tokenData) {
  * @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.
+ * @param {String} [query.email] - The email of the user.
+ * @param {String} [query.identifier] - Unique, alternative identifier for the token.
  * @returns {Promise<Object|null>} The matched Token document, or null if not found.
  * @throws Will throw an error if the find operation fails.
  */
@@ -59,6 +86,9 @@ async function findToken(query) {
     if (query.email) {
       conditions.push({ email: query.email });
     }
+    if (query.identifier) {
+      conditions.push({ identifier: query.identifier });
+    }
 
     const token = await Token.findOne({
       $and: conditions,
@@ -76,6 +106,8 @@ async function findToken(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.
+ * @param {String} [query.identifier] - Unique, alternative identifier for the token.
  * @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.
@@ -94,14 +126,20 @@ async function updateToken(query, updateData) {
  * @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.
+ * @param {String} [query.email] - The email of the user.
+ * @param {String} [query.identifier] - Unique, alternative identifier for the token.
  * @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 }],
+      $or: [
+        { userId: query.userId },
+        { token: query.token },
+        { email: query.email },
+        { identifier: query.identifier },
+      ],
     });
   } catch (error) {
     logger.debug('An error occurred while deleting tokens:', error);
@@ -109,9 +147,46 @@ async function deleteTokens(query) {
   }
 }
 
+/**
+ * Handles the OAuth token by creating or updating the token.
+ * @param {object} fields
+ * @param {string} fields.userId - The user's ID.
+ * @param {string} fields.token - The full token to store.
+ * @param {string} fields.identifier - Unique, alternative identifier for the token.
+ * @param {number} fields.expiresIn - The number of seconds until the token expires.
+ * @param {object} fields.metadata - Additional metadata to store with the token.
+ * @param {string} [fields.type="oauth"] - The type of token. Default is 'oauth'.
+ */
+async function handleOAuthToken({
+  token,
+  userId,
+  identifier,
+  expiresIn,
+  metadata,
+  type = 'oauth',
+}) {
+  const encrypedToken = await encryptV2(token);
+  const tokenData = {
+    type,
+    userId,
+    metadata,
+    identifier,
+    token: encrypedToken,
+    expiresIn: parseInt(expiresIn, 10) || 3600,
+  };
+
+  const existingToken = await findToken({ userId, identifier });
+  if (existingToken) {
+    return await updateToken({ identifier }, tokenData);
+  } else {
+    return await createToken(tokenData);
+  }
+}
+
 module.exports = {
-  createToken,
   findToken,
+  createToken,
   updateToken,
   deleteTokens,
+  handleOAuthToken,
 };

api/models/schema/agent.js

@@ -35,6 +35,9 @@ const agentSchema = mongoose.Schema(
     model_parameters: {
       type: Object,
     },
+    artifacts: {
+      type: String,
+    },
     access_level: {
       type: Number,
     },

api/models/schema/convoSchema.js

@@ -29,22 +29,6 @@ const convoSchema = mongoose.Schema(
     agent_id: {
       type: String,
     },
-    // for bingAI only
-    bingConversationId: {
-      type: String,
-    },
-    jailbreakConversationId: {
-      type: String,
-    },
-    conversationSignature: {
-      type: String,
-    },
-    clientId: {
-      type: String,
-    },
-    invocationId: {
-      type: Number,
-    },
     tags: {
       type: [String],
       default: [],
@@ -53,6 +37,9 @@ const convoSchema = mongoose.Schema(
     files: {
       type: [String],
     },
+    expiredAt: {
+      type: Date,
+    },
   },
   { timestamps: true },
 );
@@ -66,6 +53,8 @@ if (process.env.MEILI_HOST && process.env.MEILI_MASTER_KEY) {
   });
 }
 
+// Create TTL index
+convoSchema.index({ expiredAt: 1 }, { expireAfterSeconds: 0 });
 convoSchema.index({ createdAt: 1, updatedAt: 1 });
 convoSchema.index({ conversationId: 1, user: 1 }, { unique: true });
 

api/models/schema/defaults.js

@@ -1,5 +1,5 @@
 const conversationPreset = {
-  // endpoint: [azureOpenAI, openAI, bingAI, anthropic, chatGPTBrowser]
+  // endpoint: [azureOpenAI, openAI, anthropic, chatGPTBrowser]
   endpoint: {
     type: String,
     default: null,
@@ -61,19 +61,6 @@ const conversationPreset = {
     type: Number,
     required: false,
   },
-  // for bingai only
-  jailbreak: {
-    type: Boolean,
-  },
-  context: {
-    type: String,
-  },
-  systemMessage: {
-    type: String,
-  },
-  toneStyle: {
-    type: String,
-  },
   file_ids: { type: [{ type: String }], default: undefined },
   // deprecated
   resendImages: {
@@ -130,6 +117,10 @@ const conversationPreset = {
   max_tokens: {
     type: Number,
   },
+  /** omni models only */
+  reasoning_effort: {
+    type: String,
+  },
 };
 
 const agentOptions = {
@@ -179,12 +170,6 @@ const agentOptions = {
     type: Number,
     required: false,
   },
-  context: {
-    type: String,
-  },
-  systemMessage: {
-    type: String,
-  },
 };
 
 module.exports = {

api/models/schema/messageSchema.js

@@ -62,10 +62,6 @@ const messageSchema = mongoose.Schema(
       required: true,
       default: false,
     },
-    isEdited: {
-      type: Boolean,
-      default: false,
-    },
     unfinished: {
       type: Boolean,
       default: false,
@@ -138,6 +134,9 @@ const messageSchema = mongoose.Schema(
       default: undefined,
     },
     */
+    expiredAt: {
+      type: Date,
+    },
   },
   { timestamps: true },
 );
@@ -150,7 +149,7 @@ if (process.env.MEILI_HOST && process.env.MEILI_MASTER_KEY) {
     primaryKey: 'messageId',
   });
 }
-
+messageSchema.index({ expiredAt: 1 }, { expireAfterSeconds: 0 });
 messageSchema.index({ createdAt: 1 });
 messageSchema.index({ messageId: 1, user: 1 }, { unique: true });
 

api/models/schema/shareSchema.js

@@ -20,14 +20,6 @@ const shareSchema = mongoose.Schema(
       index: true,
     },
     isPublic: {
-      type: Boolean,
-      default: false,
-    },
-    isVisible: {
-      type: Boolean,
-      default: false,
-    },
-    isAnonymous: {
       type: Boolean,
       default: true,
     },

api/models/schema/tokenSchema.js

@@ -10,6 +10,10 @@ const tokenSchema = new Schema({
   email: {
     type: String,
   },
+  type: String,
+  identifier: {
+    type: String,
+  },
   token: {
     type: String,
     required: true,
@@ -23,6 +27,10 @@ const tokenSchema = new Schema({
     type: Date,
     required: true,
   },
+  metadata: {
+    type: Map,
+    of: Schema.Types.Mixed,
+  },
 });
 
 tokenSchema.index({ expiresAt: 1 }, { expireAfterSeconds: 0 });

api/models/schema/userSchema.js

@@ -23,6 +23,7 @@ const { SystemRoles } = require('librechat-data-provider');
  * @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 {string} [appleId] - Optional Apple 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
@@ -111,6 +112,11 @@ const userSchema = mongoose.Schema(
       unique: true,
       sparse: true,
     },
+    appleId: {
+      type: String,
+      unique: true,
+      sparse: true,
+    },
     plugins: {
       type: Array,
       default: [],

api/models/tx.js

@@ -75,8 +75,9 @@ const tokenValues = Object.assign(
     '4k': { prompt: 1.5, completion: 2 },
     '16k': { prompt: 3, completion: 4 },
     'gpt-3.5-turbo-1106': { prompt: 1, completion: 2 },
+    'o3-mini': { prompt: 1.1, completion: 4.4 },
+    'o1-mini': { prompt: 1.1, completion: 4.4 },
     'o1-preview': { prompt: 15, completion: 60 },
-    'o1-mini': { prompt: 3, completion: 12 },
     o1: { prompt: 15, completion: 60 },
     'gpt-4o-mini': { prompt: 0.15, completion: 0.6 },
     'gpt-4o': { prompt: 2.5, completion: 10 },
@@ -96,12 +97,19 @@ const tokenValues = Object.assign(
     'claude-': { prompt: 0.8, completion: 2.4 },
     'command-r-plus': { prompt: 3, completion: 15 },
     'command-r': { prompt: 0.5, completion: 1.5 },
+    'deepseek-reasoner': { prompt: 0.55, completion: 2.19 },
+    deepseek: { prompt: 0.14, completion: 0.28 },
     /* 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-2.0-flash-lite': { prompt: 0.075, completion: 0.3 },
+    'gemini-2.0-flash': { prompt: 0.1, completion: 0.7 },
     'gemini-2.0': { prompt: 0, completion: 0 }, // https://ai.google.dev/pricing
-    '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-flash-8b': { prompt: 0.075, completion: 0.3 },
+    'gemini-1.5-flash': { prompt: 0.15, completion: 0.6 },
+    'gemini-1.5': { prompt: 2.5, completion: 10 },
+    'gemini-pro-vision': { prompt: 0.5, completion: 1.5 },
+    gemini: { prompt: 0.5, completion: 1.5 },
   },
   bedrockValues,
 );

api/models/tx.spec.js

@@ -263,6 +263,37 @@ describe('AWS Bedrock Model Tests', () => {
   });
 });
 
+describe('Deepseek Model Tests', () => {
+  const deepseekModels = ['deepseek-chat', 'deepseek-coder', 'deepseek-reasoner'];
+
+  it('should return the correct prompt multipliers for all models', () => {
+    const results = deepseekModels.map((model) => {
+      const valueKey = getValueKey(model);
+      const multiplier = getMultiplier({ valueKey, tokenType: 'prompt' });
+      return tokenValues[valueKey].prompt && multiplier === tokenValues[valueKey].prompt;
+    });
+    expect(results.every(Boolean)).toBe(true);
+  });
+
+  it('should return the correct completion multipliers for all models', () => {
+    const results = deepseekModels.map((model) => {
+      const valueKey = getValueKey(model);
+      const multiplier = getMultiplier({ valueKey, tokenType: 'completion' });
+      return tokenValues[valueKey].completion && multiplier === tokenValues[valueKey].completion;
+    });
+    expect(results.every(Boolean)).toBe(true);
+  });
+
+  it('should return the correct prompt multipliers for reasoning model', () => {
+    const model = 'deepseek-reasoner';
+    const valueKey = getValueKey(model);
+    expect(valueKey).toBe(model);
+    const multiplier = getMultiplier({ valueKey, tokenType: 'prompt' });
+    const result = tokenValues[valueKey].prompt && multiplier === tokenValues[valueKey].prompt;
+    expect(result).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(
@@ -349,3 +380,81 @@ describe('getCacheMultiplier', () => {
     ).toBe(0.03);
   });
 });
+
+describe('Google Model Tests', () => {
+  const googleModels = [
+    'gemini-2.0-flash-lite-preview-02-05',
+    'gemini-2.0-flash-001',
+    'gemini-2.0-flash-exp',
+    'gemini-2.0-pro-exp-02-05',
+    'gemini-1.5-flash-8b',
+    'gemini-1.5-flash-thinking',
+    'gemini-1.5-pro-latest',
+    'gemini-1.5-pro-preview-0409',
+    'gemini-pro-vision',
+    'gemini-1.0',
+    'gemini-pro',
+  ];
+
+  it('should return the correct prompt and completion rates for all models', () => {
+    const results = googleModels.map((model) => {
+      const valueKey = getValueKey(model, EModelEndpoint.google);
+      const promptRate = getMultiplier({
+        model,
+        tokenType: 'prompt',
+        endpoint: EModelEndpoint.google,
+      });
+      const completionRate = getMultiplier({
+        model,
+        tokenType: 'completion',
+        endpoint: EModelEndpoint.google,
+      });
+      return { model, valueKey, promptRate, completionRate };
+    });
+
+    results.forEach(({ valueKey, promptRate, completionRate }) => {
+      expect(promptRate).toBe(tokenValues[valueKey].prompt);
+      expect(completionRate).toBe(tokenValues[valueKey].completion);
+    });
+  });
+
+  it('should map to the correct model keys', () => {
+    const expected = {
+      'gemini-2.0-flash-lite-preview-02-05': 'gemini-2.0-flash-lite',
+      'gemini-2.0-flash-001': 'gemini-2.0-flash',
+      'gemini-2.0-flash-exp': 'gemini-2.0-flash',
+      'gemini-2.0-pro-exp-02-05': 'gemini-2.0',
+      'gemini-1.5-flash-8b': 'gemini-1.5-flash-8b',
+      'gemini-1.5-flash-thinking': 'gemini-1.5-flash',
+      'gemini-1.5-pro-latest': 'gemini-1.5',
+      'gemini-1.5-pro-preview-0409': 'gemini-1.5',
+      'gemini-pro-vision': 'gemini-pro-vision',
+      'gemini-1.0': 'gemini',
+      'gemini-pro': 'gemini',
+    };
+
+    Object.entries(expected).forEach(([model, expectedKey]) => {
+      const valueKey = getValueKey(model, EModelEndpoint.google);
+      expect(valueKey).toBe(expectedKey);
+    });
+  });
+
+  it('should handle model names with different formats', () => {
+    const testCases = [
+      { input: 'google/gemini-pro', expected: 'gemini' },
+      { input: 'gemini-pro/google', expected: 'gemini' },
+      { input: 'google/gemini-2.0-flash-lite', expected: 'gemini-2.0-flash-lite' },
+    ];
+
+    testCases.forEach(({ input, expected }) => {
+      const valueKey = getValueKey(input, EModelEndpoint.google);
+      expect(valueKey).toBe(expected);
+      expect(
+        getMultiplier({ model: input, tokenType: 'prompt', endpoint: EModelEndpoint.google }),
+      ).toBe(tokenValues[expected].prompt);
+      expect(
+        getMultiplier({ model: input, tokenType: 'completion', endpoint: EModelEndpoint.google }),
+      ).toBe(tokenValues[expected].completion);
+    });
+  });
+});

api/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@librechat/backend",
-  "version": "v0.7.6",
+  "version": "v0.7.7-rc1",
   "description": "",
   "scripts": {
     "start": "echo 'please run this from the root directory'",
@@ -37,17 +37,18 @@
     "@anthropic-ai/sdk": "^0.32.1",
     "@azure/search-documents": "^12.0.0",
     "@google/generative-ai": "^0.21.0",
+    "@googleapis/youtube": "^20.0.0",
     "@keyv/mongo": "^2.1.8",
     "@keyv/redis": "^2.8.1",
     "@langchain/community": "^0.3.14",
-    "@langchain/core": "^0.3.18",
-    "@langchain/google-genai": "^0.1.6",
-    "@langchain/google-vertexai": "^0.1.6",
+    "@langchain/core": "^0.3.37",
+    "@langchain/google-genai": "^0.1.7",
+    "@langchain/google-vertexai": "^0.1.8",
     "@langchain/textsplitters": "^0.1.0",
-    "@librechat/agents": "^1.9.94",
-    "axios": "^1.7.7",
+    "@librechat/agents": "^2.0.5",
+    "@waylaidwanderer/fetch-event-source": "^3.0.1",
+    "axios": "1.7.8",
     "bcryptjs": "^2.4.3",
-    "cheerio": "^1.0.0-rc.12",
     "cohere-ai": "^7.9.1",
     "compression": "^1.7.4",
     "connect-redis": "^7.1.0",
@@ -64,7 +65,6 @@
     "firebase": "^11.0.2",
     "googleapis": "^126.0.1",
     "handlebars": "^4.7.7",
-    "html": "^1.0.0",
     "ioredis": "^5.3.2",
     "js-yaml": "^4.1.0",
     "jsonwebtoken": "^9.0.0",
@@ -82,14 +82,13 @@
     "mongoose": "^8.9.5",
     "multer": "^1.4.5-lts.1",
     "nanoid": "^3.3.7",
-    "nodejs-gpt": "^1.37.4",
     "nodemailer": "^6.9.15",
     "ollama": "^0.5.0",
     "openai": "^4.47.1",
     "openai-chat-tokens": "^0.2.8",
     "openid-client": "^5.4.2",
     "passport": "^0.6.0",
-    "passport-custom": "^1.1.1",
+    "passport-apple": "^2.0.2",
     "passport-discord": "^0.1.4",
     "passport-facebook": "^3.0.0",
     "passport-github2": "^0.1.12",
@@ -97,19 +96,21 @@
     "passport-jwt": "^4.0.1",
     "passport-ldapauth": "^3.0.1",
     "passport-local": "^1.0.0",
-    "pino": "^8.12.1",
     "sharp": "^0.32.6",
+    "socket.io": "^4.8.1",
     "tiktoken": "^1.0.15",
     "traverse": "^0.6.7",
     "ua-parser-js": "^1.0.36",
     "winston": "^3.11.0",
     "winston-daily-rotate-file": "^4.7.1",
+    "wrtc": "^0.4.7",
+    "youtube-transcript": "^1.2.1",
     "zod": "^3.22.4"
   },
   "devDependencies": {
     "jest": "^29.7.0",
-    "mongodb-memory-server": "^10.0.0",
-    "nodemon": "^3.0.1",
-    "supertest": "^6.3.3"
+    "mongodb-memory-server": "^10.1.3",
+    "nodemon": "^3.0.3",
+    "supertest": "^7.0.0"
   }
 }

api/server/controllers/AskController.js

@@ -1,8 +1,6 @@
-const throttle = require('lodash/throttle');
-const { getResponseSender, Constants, CacheKeys, Time } = require('librechat-data-provider');
+const { getResponseSender, Constants } = require('librechat-data-provider');
 const { createAbortController, handleAbortError } = require('~/server/middleware');
 const { sendMessage, createOnProgress } = require('~/server/utils');
-const { getLogStores } = require('~/cache');
 const { saveMessage } = require('~/models');
 const { logger } = require('~/config');
 
@@ -57,33 +55,9 @@ const AskController = async (req, res, next, initializeClient, addTitle) => {
 
   try {
     const { client } = await initializeClient({ req, res, endpointOption });
-    const messageCache = getLogStores(CacheKeys.MESSAGES);
-    const { onProgress: progressCallback, getPartialText } = createOnProgress({
-      onProgress: throttle(
-        ({ text: partialText }) => {
-          /*
-              const unfinished = endpointOption.endpoint === EModelEndpoint.google ? false : true;
-          messageCache.set(responseMessageId, {
-            messageId: responseMessageId,
-            sender,
-            conversationId,
-            parentMessageId: overrideParentMessageId ?? userMessageId,
-            text: partialText,
-            model: client.modelOptions.model,
-            unfinished,
-            error: false,
-            user,
-          }, Time.FIVE_MINUTES);
-          */
+    const { onProgress: progressCallback, getPartialText } = createOnProgress();
 
-          messageCache.set(responseMessageId, partialText, Time.FIVE_MINUTES);
-        },
-        3000,
-        { trailing: false },
-      ),
-    });
-
-    getText = getPartialText;
+    getText = client.getStreamText != null ? client.getStreamText.bind(client) : getPartialText;
 
     const getAbortData = () => ({
       sender,
@@ -91,7 +65,7 @@ const AskController = async (req, res, next, initializeClient, addTitle) => {
       userMessagePromise,
       messageId: responseMessageId,
       parentMessageId: overrideParentMessageId ?? userMessageId,
-      text: getPartialText(),
+      text: getText(),
       userMessage,
       promptTokens,
     });
@@ -176,11 +150,13 @@ const AskController = async (req, res, next, initializeClient, addTitle) => {
   } catch (error) {
     const partialText = getText && getText();
     handleAbortError(res, req, error, {
+      sender,
       partialText,
       conversationId,
-      sender,
       messageId: responseMessageId,
-      parentMessageId: userMessageId ?? parentMessageId,
+      parentMessageId: overrideParentMessageId ?? userMessageId ?? parentMessageId,
+    }).catch((err) => {
+      logger.error('[AskController] Error in `handleAbortError`', err);
     });
   }
 };

api/server/controllers/EditController.js

@@ -1,8 +1,6 @@
-const throttle = require('lodash/throttle');
-const { getResponseSender, CacheKeys, Time } = require('librechat-data-provider');
+const { getResponseSender } = require('librechat-data-provider');
 const { createAbortController, handleAbortError } = require('~/server/middleware');
 const { sendMessage, createOnProgress } = require('~/server/utils');
-const { getLogStores } = require('~/cache');
 const { saveMessage } = require('~/models');
 const { logger } = require('~/config');
 
@@ -53,62 +51,44 @@ const EditController = async (req, res, next, initializeClient) => {
     }
   };
 
-  const messageCache = getLogStores(CacheKeys.MESSAGES);
   const { onProgress: progressCallback, getPartialText } = createOnProgress({
     generation,
-    onProgress: throttle(
-      ({ text: partialText }) => {
-        /*
-          const unfinished = endpointOption.endpoint === EModelEndpoint.google ? false : true;
-        {
-          messageId: responseMessageId,
-          sender,
-          conversationId,
-          parentMessageId: overrideParentMessageId ?? userMessageId,
-          text: partialText,
-          model: endpointOption.modelOptions.model,
-          unfinished,
-          isEdited: true,
-          error: false,
-          user,
-        } */
-        messageCache.set(responseMessageId, partialText, Time.FIVE_MINUTES);
-      },
-      3000,
-      { trailing: false },
-    ),
   });
 
-  const getAbortData = () => ({
-    conversationId,
-    userMessagePromise,
-    messageId: responseMessageId,
-    sender,
-    parentMessageId: overrideParentMessageId ?? userMessageId,
-    text: getPartialText(),
-    userMessage,
-    promptTokens,
-  });
-
-  const { abortController, onStart } = createAbortController(req, res, getAbortData, getReqData);
-
-  res.on('close', () => {
-    logger.debug('[EditController] Request closed');
-    if (!abortController) {
-      return;
-    } else if (abortController.signal.aborted) {
-      return;
-    } else if (abortController.requestCompleted) {
-      return;
-    }
-
-    abortController.abort();
-    logger.debug('[EditController] Request aborted on close');
-  });
+  let getText;
 
   try {
     const { client } = await initializeClient({ req, res, endpointOption });
 
+    getText = client.getStreamText != null ? client.getStreamText.bind(client) : getPartialText;
+
+    const getAbortData = () => ({
+      conversationId,
+      userMessagePromise,
+      messageId: responseMessageId,
+      sender,
+      parentMessageId: overrideParentMessageId ?? userMessageId,
+      text: getText(),
+      userMessage,
+      promptTokens,
+    });
+
+    const { abortController, onStart } = createAbortController(req, res, getAbortData, getReqData);
+
+    res.on('close', () => {
+      logger.debug('[EditController] Request closed');
+      if (!abortController) {
+        return;
+      } else if (abortController.signal.aborted) {
+        return;
+      } else if (abortController.requestCompleted) {
+        return;
+      }
+
+      abortController.abort();
+      logger.debug('[EditController] Request aborted on close');
+    });
+
     let response = await client.sendMessage(text, {
       user,
       generation,
@@ -153,13 +133,15 @@ const EditController = async (req, res, next, initializeClient) => {
       );
     }
   } catch (error) {
-    const partialText = getPartialText();
+    const partialText = getText();
     handleAbortError(res, req, error, {
+      sender,
       partialText,
       conversationId,
-      sender,
       messageId: responseMessageId,
-      parentMessageId: userMessageId ?? parentMessageId,
+      parentMessageId: overrideParentMessageId ?? userMessageId ?? parentMessageId,
+    }).catch((err) => {
+      logger.error('[EditController] Error in `handleAbortError`', err);
     });
   }
 };

api/server/controllers/PluginController.js

@@ -1,7 +1,7 @@
-const { promises: fs } = require('fs');
 const { CacheKeys, AuthType } = require('librechat-data-provider');
 const { addOpenAPISpecs } = require('~/app/clients/tools/util/addOpenAPISpecs');
 const { getCustomConfig } = require('~/server/services/Config');
+const { availableTools } = require('~/app/clients/tools');
 const { getMCPManager } = require('~/config');
 const { getLogStores } = require('~/cache');
 
@@ -59,10 +59,9 @@ const getAvailablePluginsController = async (req, res) => {
 
     /** @type {{ filteredTools: string[], includedTools: string[] }} */
     const { filteredTools = [], includedTools = [] } = req.app.locals;
-    const pluginManifest = await fs.readFile(req.app.locals.paths.pluginManifest, 'utf8');
-    const jsonData = JSON.parse(pluginManifest);
+    const pluginManifest = availableTools;
 
-    const uniquePlugins = filterUniquePlugins(jsonData);
+    const uniquePlugins = filterUniquePlugins(pluginManifest);
     let authenticatedPlugins = [];
     for (const plugin of uniquePlugins) {
       authenticatedPlugins.push(
@@ -106,17 +105,15 @@ const getAvailableTools = async (req, res) => {
       return;
     }
 
-    const pluginManifest = await fs.readFile(req.app.locals.paths.pluginManifest, 'utf8');
-
-    const jsonData = JSON.parse(pluginManifest);
+    const pluginManifest = availableTools;
     const customConfig = await getCustomConfig();
     if (customConfig?.mcpServers != null) {
       const mcpManager = await getMCPManager();
-      await mcpManager.loadManifestTools(jsonData);
+      await mcpManager.loadManifestTools(pluginManifest);
     }
 
     /** @type {TPlugin[]} */
-    const uniquePlugins = filterUniquePlugins(jsonData);
+    const uniquePlugins = filterUniquePlugins(pluginManifest);
 
     const authenticatedPlugins = uniquePlugins.map((plugin) => {
       if (checkPluginAuth(plugin)) {
@@ -126,8 +123,12 @@ const getAvailableTools = async (req, res) => {
       }
     });
 
+    const toolDefinitions = req.app.locals.availableTools;
     const tools = authenticatedPlugins.filter(
-      (plugin) => req.app.locals.availableTools[plugin.pluginKey] !== undefined,
+      (plugin) =>
+        toolDefinitions[plugin.pluginKey] !== undefined ||
+        (plugin.toolkit === true &&
+          Object.keys(toolDefinitions).some((key) => key.startsWith(`${plugin.pluginKey}_`))),
     );
 
     await cache.set(CacheKeys.TOOLS, tools);

api/server/controllers/agents/callbacks.js

@@ -3,6 +3,7 @@ const {
   EnvVar,
   Providers,
   GraphEvents,
+  getMessageId,
   ToolEndHandler,
   handleToolCalls,
   ChatModelStreamHandler,
@@ -10,7 +11,7 @@ const {
 const { processCodeOutput } = require('~/server/services/Files/Code/process');
 const { saveBase64Image } = require('~/server/services/Files/process');
 const { loadAuthValues } = require('~/app/clients/tools/util');
-const { logger } = require('~/config');
+const { logger, sendEvent } = require('~/config');
 
 /** @typedef {import('@librechat/agents').Graph} Graph */
 /** @typedef {import('@librechat/agents').EventHandler} EventHandler */
@@ -21,20 +22,6 @@ const { logger } = require('~/config');
 /** @typedef {import('@librechat/agents').ContentAggregatorResult['aggregateContent']} ContentAggregator */
 /** @typedef {import('@librechat/agents').GraphEvents} GraphEvents */
 
-/**
- * Sends message data in Server Sent Events format.
- * @param {ServerResponse} res - The server response.
- * @param {{ data: string | Record<string, unknown>, event?: string }} event - The message event.
- * @param {string} event.event - The type of event.
- * @param {string} event.data - The message to be sent.
- */
-const sendEvent = (res, event) => {
-  if (typeof event.data === 'string' && event.data.length === 0) {
-    return;
-  }
-  res.write(`event: message\ndata: ${JSON.stringify(event)}\n\n`);
-};
-
 class ModelEndHandler {
   /**
    * @param {Array<UsageMetadata>} collectedUsage
@@ -60,7 +47,7 @@ class ModelEndHandler {
     }
 
     try {
-      if (metadata.provider === Providers.GOOGLE) {
+      if (metadata.provider === Providers.GOOGLE || graph.clientOptions?.disableStreaming) {
         handleToolCalls(data?.output?.tool_calls, metadata, graph);
       }
 
@@ -73,6 +60,38 @@ class ModelEndHandler {
       }
 
       this.collectedUsage.push(usage);
+      if (!graph.clientOptions?.disableStreaming) {
+        return;
+      }
+      if (!data.output.content) {
+        return;
+      }
+      const stepKey = graph.getStepKey(metadata);
+      const message_id = getMessageId(stepKey, graph) ?? '';
+      if (message_id) {
+        graph.dispatchRunStep(stepKey, {
+          type: StepTypes.MESSAGE_CREATION,
+          message_creation: {
+            message_id,
+          },
+        });
+      }
+      const stepId = graph.getStepIdByKey(stepKey);
+      const content = data.output.content;
+      if (typeof content === 'string') {
+        graph.dispatchMessageDelta(stepId, {
+          content: [
+            {
+              type: 'text',
+              text: content,
+            },
+          ],
+        });
+      } else if (content.every((c) => c.type?.startsWith('text'))) {
+        graph.dispatchMessageDelta(stepId, {
+          content,
+        });
+      }
     } catch (error) {
       logger.error('Error handling model end event:', error);
     }
@@ -322,7 +341,6 @@ function createToolEndCallback({ req, res, artifactPromises }) {
 }
 
 module.exports = {
-  sendEvent,
   getDefaultHandlers,
   createToolEndCallback,
 };

api/server/controllers/agents/request.js

@@ -142,7 +142,9 @@ const AgentController = async (req, res, next, initializeClient, addTitle) => {
       conversationId,
       sender,
       messageId: responseMessageId,
-      parentMessageId: userMessageId ?? parentMessageId,
+      parentMessageId: overrideParentMessageId ?? userMessageId ?? parentMessageId,
+    }).catch((err) => {
+      logger.error('[api/server/controllers/agents/request] Error in `handleAbortError`', err);
     });
   }
 };

api/server/controllers/agents/run.js

@@ -41,6 +41,11 @@ async function createRun({
     agent.model_parameters,
   );
 
+  if (/o1(?!-(?:mini|preview)).*$/.test(llmConfig.model)) {
+    llmConfig.streaming = false;
+    llmConfig.disableStreaming = true;
+  }
+
   /** @type {StandardGraphConfig} */
   const graphConfig = {
     signal,

api/server/controllers/assistants/chatV2.js

@@ -397,18 +397,6 @@ const chatV2 = async (req, res) => {
 
       response = streamRunManager;
       response.text = streamRunManager.intermediateText;
-
-      if (response.text) {
-        const messageCache = getLogStores(CacheKeys.MESSAGES);
-        messageCache.set(
-          responseMessageId,
-          {
-            complete: true,
-            text: response.text,
-          },
-          Time.FIVE_MINUTES,
-        );
-      }
     };
 
     await processRun();

api/server/controllers/assistants/v1.js

@@ -6,6 +6,7 @@ const { getStrategyFunctions } = require('~/server/services/Files/strategies');
 const { deleteAssistantActions } = require('~/server/services/ActionService');
 const { updateAssistantDoc, getAssistants } = require('~/models/Assistant');
 const { getOpenAIClient, fetchAssistants } = require('./helpers');
+const { manifestToolMap } = require('~/app/clients/tools');
 const { deleteFileByFilter } = require('~/models/File');
 const { logger } = require('~/config');
 
@@ -35,9 +36,21 @@ const createAssistant = async (req, res) => {
           return tool;
         }
 
-        return req.app.locals.availableTools[tool];
+        const toolDefinitions = req.app.locals.availableTools;
+        const toolDef = toolDefinitions[tool];
+        if (!toolDef && manifestToolMap[tool] && manifestToolMap[tool].toolkit === true) {
+          return (
+            Object.entries(toolDefinitions)
+              .filter(([key]) => key.startsWith(`${tool}_`))
+              // eslint-disable-next-line no-unused-vars
+              .map(([_, val]) => val)
+          );
+        }
+
+        return toolDef;
       })
-      .filter((tool) => tool);
+      .filter((tool) => tool)
+      .flat();
 
     let azureModelIdentifier = null;
     if (openai.locals?.azureOptions) {
@@ -128,9 +141,21 @@ const patchAssistant = async (req, res) => {
           return tool;
         }
 
-        return req.app.locals.availableTools[tool];
+        const toolDefinitions = req.app.locals.availableTools;
+        const toolDef = toolDefinitions[tool];
+        if (!toolDef && manifestToolMap[tool] && manifestToolMap[tool].toolkit === true) {
+          return (
+            Object.entries(toolDefinitions)
+              .filter(([key]) => key.startsWith(`${tool}_`))
+              // eslint-disable-next-line no-unused-vars
+              .map(([_, val]) => val)
+          );
+        }
+
+        return toolDef;
       })
-      .filter((tool) => tool);
+      .filter((tool) => tool)
+      .flat();
 
     if (openai.locals?.azureOptions && updateData.model) {
       updateData.model = openai.locals.azureOptions.azureOpenAIApiDeploymentName;

api/server/controllers/assistants/v2.js

@@ -2,6 +2,7 @@ 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 { manifestToolMap } = require('~/app/clients/tools');
 const { getOpenAIClient } = require('./helpers');
 const { logger } = require('~/config');
 
@@ -32,9 +33,21 @@ const createAssistant = async (req, res) => {
           return tool;
         }
 
-        return req.app.locals.availableTools[tool];
+        const toolDefinitions = req.app.locals.availableTools;
+        const toolDef = toolDefinitions[tool];
+        if (!toolDef && manifestToolMap[tool] && manifestToolMap[tool].toolkit === true) {
+          return (
+            Object.entries(toolDefinitions)
+              .filter(([key]) => key.startsWith(`${tool}_`))
+              // eslint-disable-next-line no-unused-vars
+              .map(([_, val]) => val)
+          );
+        }
+
+        return toolDef;
       })
-      .filter((tool) => tool);
+      .filter((tool) => tool)
+      .flat();
 
     let azureModelIdentifier = null;
     if (openai.locals?.azureOptions) {
@@ -112,9 +125,30 @@ const updateAssistant = async ({ req, openai, assistant_id, updateData }) => {
 
   let hasFileSearch = false;
   for (const tool of updateData.tools ?? []) {
-    let actualTool = typeof tool === 'string' ? req.app.locals.availableTools[tool] : tool;
+    const toolDefinitions = req.app.locals.availableTools;
+    let actualTool = typeof tool === 'string' ? toolDefinitions[tool] : tool;
 
-    if (!actualTool) {
+    if (!actualTool && manifestToolMap[tool] && manifestToolMap[tool].toolkit === true) {
+      actualTool = Object.entries(toolDefinitions)
+        .filter(([key]) => key.startsWith(`${tool}_`))
+        // eslint-disable-next-line no-unused-vars
+        .map(([_, val]) => val);
+    } else if (!actualTool) {
+      continue;
+    }
+
+    if (Array.isArray(actualTool)) {
+      for (const subTool of actualTool) {
+        if (!subTool.function) {
+          tools.push(subTool);
+          continue;
+        }
+
+        const updatedTool = await validateAndUpdateTool({ req, tool: subTool, assistant_id });
+        if (updatedTool) {
+          tools.push(updatedTool);
+        }
+      }
       continue;
     }
 

api/server/controllers/auth/LogoutController.js

@@ -1,14 +1,32 @@
 const cookies = require('cookie');
+const { Issuer } = require('openid-client');
 const { logoutUser } = require('~/server/services/AuthService');
+const { isEnabled } = require('~/server/utils');
 const { logger } = require('~/config');
 
 const logoutController = async (req, res) => {
   const refreshToken = req.headers.cookie ? cookies.parse(req.headers.cookie).refreshToken : null;
   try {
-    const logout = await logoutUser(req.user._id, refreshToken);
+    const logout = await logoutUser(req, refreshToken);
     const { status, message } = logout;
     res.clearCookie('refreshToken');
-    return res.status(status).send({ message });
+    const response = { message };
+    if (
+      req.user.openidId != null &&
+      isEnabled(process.env.OPENID_USE_END_SESSION_ENDPOINT) &&
+      process.env.OPENID_ISSUER
+    ) {
+      const issuer = await Issuer.discover(process.env.OPENID_ISSUER);
+      const redirect = issuer.metadata.end_session_endpoint;
+      if (!redirect) {
+        logger.warn(
+          '[logoutController] end_session_endpoint not found in OpenID issuer metadata. Please verify that the issuer is correct.',
+        );
+      } else {
+        response.redirect = redirect;
+      }
+    }
+    return res.status(status).send(response);
   } catch (err) {
     logger.error('[logoutController]', err);
     return res.status(500).json({ message: err.message });

api/server/index.js

@@ -4,6 +4,7 @@ require('module-alias')({ base: path.resolve(__dirname, '..') });
 const cors = require('cors');
 const axios = require('axios');
 const express = require('express');
+const { createServer } = require('http');
 const compression = require('compression');
 const passport = require('passport');
 const mongoSanitize = require('express-mongo-sanitize');
@@ -14,6 +15,7 @@ const { connectDb, indexSync } = require('~/lib/db');
 const { isEnabled } = require('~/server/utils');
 const { ldapLogin } = require('~/strategies');
 const { logger } = require('~/config');
+const { SocketIOService } = require('./services/WebSocket/WebSocketServer');
 const validateImageRequest = require('./middleware/validateImageRequest');
 const errorController = require('./controllers/ErrorController');
 const configureSocialLogins = require('./socialLogins');
@@ -36,7 +38,18 @@ const startServer = async () => {
   await indexSync();
 
   const app = express();
+  const server = createServer(app);
+
   app.disable('x-powered-by');
+  app.use(
+    cors({
+      origin: true,
+      credentials: true,
+    }),
+  );
+
+  new SocketIOService(server);
+
   await AppService(app);
 
   const indexPath = path.join(app.locals.paths.dist, 'index.html');
@@ -84,6 +97,7 @@ const startServer = async () => {
   app.use('/oauth', routes.oauth);
   /* API Endpoints */
   app.use('/api/auth', routes.auth);
+  app.use('/api/actions', routes.actions);
   app.use('/api/keys', routes.keys);
   app.use('/api/user', routes.user);
   app.use('/api/search', routes.search);
@@ -108,6 +122,7 @@ const startServer = async () => {
   app.use('/api/agents', routes.agents);
   app.use('/api/banner', routes.banner);
   app.use('/api/bedrock', routes.bedrock);
+  app.use('/api/websocket', routes.websocket);
 
   app.use('/api/tags', routes.tags);
 
@@ -125,7 +140,7 @@ const startServer = async () => {
     res.send(updatedIndexHtml);
   });
 
-  app.listen(port, host, () => {
+  server.listen(port, host, () => {
     if (host == '0.0.0.0') {
       logger.info(
         `Server listening on all interfaces at port ${port}. Use http://localhost:${port} to access it`,
@@ -133,6 +148,8 @@ const startServer = async () => {
     } else {
       logger.info(`Server listening at http://${host == '0.0.0.0' ? 'localhost' : host}:${port}`);
     }
+
+    logger.info(`Socket.IO endpoint: http://${host}:${port}`);
   });
 };
 

api/server/middleware/abortMiddleware.js

@@ -75,8 +75,9 @@ const createAbortController = (req, res, getAbortData, getReqData) => {
 
     const abortKey = userMessage?.conversationId ?? req.user.id;
     const prevRequest = abortControllers.get(abortKey);
+    const { overrideUserMessageId } = req?.body ?? {};
 
-    if (prevRequest && prevRequest?.abortController) {
+    if (overrideUserMessageId != null && prevRequest && prevRequest?.abortController) {
       const data = prevRequest.abortController.getAbortData();
       getReqData({ userMessage: data?.userMessage });
       const addedAbortKey = `${abortKey}:${responseMessageId}`;

api/server/routes/actions.js

@@ -0,0 +1,136 @@
+const express = require('express');
+const jwt = require('jsonwebtoken');
+const { getAccessToken } = require('~/server/services/TokenService');
+const { logger, getFlowStateManager } = require('~/config');
+const { getLogStores } = require('~/cache');
+
+const router = express.Router();
+const JWT_SECRET = process.env.JWT_SECRET;
+
+/**
+ * Handles the OAuth callback and exchanges the authorization code for tokens.
+ *
+ * @route GET /actions/:action_id/oauth/callback
+ * @param {string} req.params.action_id - The ID of the action.
+ * @param {string} req.query.code - The authorization code returned by the provider.
+ * @param {string} req.query.state - The state token to verify the authenticity of the request.
+ * @returns {void} Sends a success message after updating the action with OAuth tokens.
+ */
+router.get('/:action_id/oauth/callback', async (req, res) => {
+  const { action_id } = req.params;
+  const { code, state } = req.query;
+
+  const flowManager = await getFlowStateManager(getLogStores);
+  let identifier = action_id;
+  try {
+    let decodedState;
+    try {
+      decodedState = jwt.verify(state, JWT_SECRET);
+    } catch (err) {
+      await flowManager.failFlow(identifier, 'oauth', 'Invalid or expired state parameter');
+      return res.status(400).send('Invalid or expired state parameter');
+    }
+
+    if (decodedState.action_id !== action_id) {
+      await flowManager.failFlow(identifier, 'oauth', 'Mismatched action ID in state parameter');
+      return res.status(400).send('Mismatched action ID in state parameter');
+    }
+
+    if (!decodedState.user) {
+      await flowManager.failFlow(identifier, 'oauth', 'Invalid user ID in state parameter');
+      return res.status(400).send('Invalid user ID in state parameter');
+    }
+    identifier = `${decodedState.user}:${action_id}`;
+    const flowState = await flowManager.getFlowState(identifier, 'oauth');
+    if (!flowState) {
+      throw new Error('OAuth flow not found');
+    }
+
+    const tokenData = await getAccessToken({
+      code,
+      userId: decodedState.user,
+      identifier,
+      client_url: flowState.metadata.client_url,
+      redirect_uri: flowState.metadata.redirect_uri,
+      /** Encrypted values */
+      encrypted_oauth_client_id: flowState.metadata.encrypted_oauth_client_id,
+      encrypted_oauth_client_secret: flowState.metadata.encrypted_oauth_client_secret,
+    });
+    await flowManager.completeFlow(identifier, 'oauth', tokenData);
+    res.send(`
+      <!DOCTYPE html>
+      <html>
+        <head>
+          <title>Authentication Successful</title>
+          <meta name="viewport" content="width=device-width, initial-scale=1.0">
+          <meta http-equiv="Content-Type" content="text/html; charset=UTF-8">
+          <style>
+            body {
+              font-family: ui-sans-serif, system-ui, -apple-system, BlinkMacSystemFont;
+              background-color: rgb(249, 250, 251);
+              margin: 0;
+              padding: 2rem;
+              display: flex;
+              justify-content: center;
+              align-items: center;
+              min-height: 100vh;
+            }
+            .card {
+              background-color: white;
+              border-radius: 0.5rem;
+              padding: 2rem;
+              max-width: 28rem;
+              width: 100%;
+              box-shadow: 0 4px 6px -1px rgb(0 0 0 / 0.1), 0 2px 4px -2px rgb(0 0 0 / 0.1);
+              text-align: center;
+            }
+            .heading {
+              color: rgb(17, 24, 39);
+              font-size: 1.875rem;
+              font-weight: 700;
+              margin: 0 0 1rem;
+            }
+            .description {
+              color: rgb(75, 85, 99);
+              font-size: 0.875rem;
+              margin: 0.5rem 0;
+            }
+            .countdown {
+              color: rgb(99, 102, 241);
+              font-weight: 500;
+            }
+          </style>
+        </head>
+        <body>
+          <div class="card">
+            <h1 class="heading">Authentication Successful</h1>
+            <p class="description">
+              Your authentication was successful. This window will close in 
+              <span class="countdown" id="countdown">3</span> seconds.
+            </p>
+          </div>
+          <script>
+            let secondsLeft = 3;
+            const countdownElement = document.getElementById('countdown');
+            
+            const countdown = setInterval(() => {
+              secondsLeft--;
+              countdownElement.textContent = secondsLeft;
+              
+              if (secondsLeft <= 0) {
+                clearInterval(countdown);
+                window.close();
+              }
+            }, 1000);
+          </script>
+        </body>
+      </html>
+    `);
+  } catch (error) {
+    logger.error('Error in OAuth callback:', error);
+    await flowManager.failFlow(identifier, 'oauth', error);
+    res.status(500).send('Authentication failed. Please try again.');
+  }
+});
+
+module.exports = router;

api/server/routes/agents/actions.js

@@ -1,6 +1,6 @@
 const express = require('express');
 const { nanoid } = require('nanoid');
-const { actionDelimiter } = require('librechat-data-provider');
+const { actionDelimiter, SystemRoles, removeNullishValues } = require('librechat-data-provider');
 const { encryptMetadata, domainParser } = require('~/server/services/ActionService');
 const { updateAction, getActions, deleteAction } = require('~/models/Action');
 const { isActionDomainAllowed } = require('~/server/services/domains');
@@ -9,6 +9,12 @@ const { logger } = require('~/config');
 
 const router = express.Router();
 
+// If the user has ADMIN role
+// then action edition is possible even if not owner of the assistant
+const isAdmin = (req) => {
+  return req.user.role === SystemRoles.ADMIN;
+};
+
 /**
  * Retrieves all user's actions
  * @route GET /actions/
@@ -17,7 +23,10 @@ const router = express.Router();
  */
 router.get('/', async (req, res) => {
   try {
-    res.json(await getActions({ user: req.user.id }));
+    const admin = isAdmin(req);
+    // If admin, get all actions, otherwise only user's actions
+    const searchParams = admin ? {} : { user: req.user.id };
+    res.json(await getActions(searchParams));
   } catch (error) {
     res.status(500).json({ error: error.message });
   }
@@ -42,7 +51,7 @@ router.post('/:agent_id', async (req, res) => {
       return res.status(400).json({ message: 'No functions provided' });
     }
 
-    let metadata = await encryptMetadata(_metadata);
+    let metadata = await encryptMetadata(removeNullishValues(_metadata, true));
     const isDomainAllowed = await isActionDomainAllowed(metadata.domain);
     if (!isDomainAllowed) {
       return res.status(400).json({ message: 'Domain not allowed' });
@@ -57,9 +66,12 @@ router.post('/:agent_id', async (req, res) => {
 
     const action_id = _action_id ?? nanoid();
     const initialPromises = [];
+    const admin = isAdmin(req);
 
+    // If admin, can edit any agent, otherwise only user's agents
+    const agentQuery = admin ? { id: agent_id } : { id: agent_id, author: req.user.id };
     // TODO: share agents
-    initialPromises.push(getAgent({ id: agent_id, author: req.user.id }));
+    initialPromises.push(getAgent(agentQuery));
     if (_action_id) {
       initialPromises.push(getActions({ action_id }, true));
     }
@@ -75,7 +87,7 @@ router.post('/:agent_id', async (req, res) => {
       metadata = { ...action.metadata, ...metadata };
     }
 
-    const { actions: _actions = [] } = agent ?? {};
+    const { actions: _actions = [], author: agent_author } = agent ?? {};
     const actions = [];
     for (const action of _actions) {
       const [_action_domain, current_action_id] = action.split(actionDelimiter);
@@ -95,15 +107,17 @@ router.post('/:agent_id', async (req, res) => {
       .filter((tool) => !(tool && (tool.includes(domain) || tool.includes(action_id))))
       .concat(functions.map((tool) => `${tool.function.name}${actionDelimiter}${domain}`));
 
-    const updatedAgent = await updateAgent(
-      { id: agent_id, author: req.user.id },
-      { tools, actions },
-    );
+    const updatedAgent = await updateAgent(agentQuery, { tools, actions });
+
+    // Only update user field for new actions
+    const actionUpdateData = { metadata, agent_id };
+    if (!actions_result || !actions_result.length) {
+      // For new actions, use the agent owner's user ID
+      actionUpdateData.user = agent_author || req.user.id;
+    }
+
     /** @type {[Action]} */
-    const updatedAction = await updateAction(
-      { action_id },
-      { metadata, agent_id, user: req.user.id },
-    );
+    const updatedAction = await updateAction({ action_id }, actionUpdateData);
 
     const sensitiveFields = ['api_key', 'oauth_client_id', 'oauth_client_secret'];
     for (let field of sensitiveFields) {
@@ -130,8 +144,11 @@ router.post('/:agent_id', async (req, res) => {
 router.delete('/:agent_id/:action_id', async (req, res) => {
   try {
     const { agent_id, action_id } = req.params;
+    const admin = isAdmin(req);
 
-    const agent = await getAgent({ id: agent_id, author: req.user.id });
+    // If admin, can delete any agent, otherwise only user's agents
+    const agentQuery = admin ? { id: agent_id } : { id: agent_id, author: req.user.id };
+    const agent = await getAgent(agentQuery);
     if (!agent) {
       return res.status(404).json({ message: 'Agent not found for deleting action' });
     }
@@ -155,11 +172,10 @@ router.delete('/:agent_id/:action_id', async (req, res) => {
 
     const updatedTools = tools.filter((tool) => !(tool && tool.includes(domain)));
 
-    await updateAgent(
-      { id: agent_id, author: req.user.id },
-      { tools: updatedTools, actions: updatedActions },
-    );
-    await deleteAction({ action_id });
+    await updateAgent(agentQuery, { tools: updatedTools, actions: updatedActions });
+    // If admin, can delete any action, otherwise only user's actions
+    const actionQuery = admin ? { action_id } : { action_id, user: req.user.id };
+    await deleteAction(actionQuery);
     res.status(200).json({ message: 'Action deleted successfully' });
   } catch (error) {
     const message = 'Trouble deleting the Agent Action';

api/server/routes/ask/addToCache.js

@@ -35,8 +35,6 @@ const addToCache = async ({ endpoint, endpointOption, userMessage, responseMessa
     const roles = (options) => {
       if (endpoint === 'openAI') {
         return options?.chatGptLabel || 'ChatGPT';
-      } else if (endpoint === 'bingAI') {
-        return options?.jailbreak ? 'Sydney' : 'BingAI';
       }
     };
 

api/server/routes/ask/askChatGPTBrowser.js

@@ -1,237 +0,0 @@
-const crypto = require('crypto');
-const express = require('express');
-const { Constants } = require('librechat-data-provider');
-const { saveMessage, getConvoTitle, saveConvo, getConvo } = require('~/models');
-const { handleError, sendMessage, createOnProgress, handleText } = require('~/server/utils');
-const { setHeaders } = require('~/server/middleware');
-const { browserClient } = require('~/app/');
-const { logger } = require('~/config');
-
-const router = express.Router();
-
-router.post('/', setHeaders, async (req, res) => {
-  const {
-    endpoint,
-    text,
-    overrideParentMessageId = null,
-    parentMessageId,
-    conversationId: oldConversationId,
-  } = req.body;
-  if (text.length === 0) {
-    return handleError(res, { text: 'Prompt empty or too short' });
-  }
-  if (endpoint !== 'chatGPTBrowser') {
-    return handleError(res, { text: 'Illegal request' });
-  }
-
-  // build user message
-  const conversationId = oldConversationId || crypto.randomUUID();
-  const isNewConversation = !oldConversationId;
-  const userMessageId = crypto.randomUUID();
-  const userParentMessageId = parentMessageId || Constants.NO_PARENT;
-  const userMessage = {
-    messageId: userMessageId,
-    sender: 'User',
-    text,
-    parentMessageId: userParentMessageId,
-    conversationId,
-    isCreatedByUser: true,
-  };
-
-  // build endpoint option
-  const endpointOption = {
-    model: req.body?.model ?? 'text-davinci-002-render-sha',
-    key: req.body?.key ?? null,
-  };
-
-  logger.debug('[/ask/chatGPTBrowser]', {
-    userMessage,
-    conversationId,
-    ...endpointOption,
-  });
-
-  if (!overrideParentMessageId) {
-    await saveMessage(req, { ...userMessage, user: req.user.id });
-    await saveConvo(req, {
-      ...userMessage,
-      ...endpointOption,
-      conversationId,
-      endpoint,
-    });
-  }
-
-  // eslint-disable-next-line no-use-before-define
-  return await ask({
-    isNewConversation,
-    userMessage,
-    endpointOption,
-    conversationId,
-    preSendRequest: true,
-    overrideParentMessageId,
-    req,
-    res,
-  });
-});
-
-const ask = async ({
-  isNewConversation,
-  userMessage,
-  endpointOption,
-  conversationId,
-  overrideParentMessageId = null,
-  req,
-  res,
-}) => {
-  let { text, parentMessageId: userParentMessageId, messageId: userMessageId } = userMessage;
-  const user = req.user.id;
-  let responseMessageId = crypto.randomUUID();
-  let getPartialMessage = null;
-  try {
-    let lastSavedTimestamp = 0;
-    const { onProgress: progressCallback, getPartialText } = createOnProgress({
-      onProgress: ({ text }) => {
-        const currentTimestamp = Date.now();
-        if (currentTimestamp - lastSavedTimestamp > 500) {
-          lastSavedTimestamp = currentTimestamp;
-          saveMessage(req, {
-            messageId: responseMessageId,
-            sender: endpointOption?.jailbreak ? 'Sydney' : 'BingAI',
-            conversationId,
-            parentMessageId: overrideParentMessageId || userMessageId,
-            text: text,
-            unfinished: true,
-            error: false,
-            isCreatedByUser: false,
-            user,
-          });
-        }
-      },
-    });
-
-    getPartialMessage = getPartialText;
-    const abortController = new AbortController();
-    let i = 0;
-    let response = await browserClient({
-      text,
-      parentMessageId: userParentMessageId,
-      conversationId,
-      ...endpointOption,
-      abortController,
-      userId: user,
-      onProgress: progressCallback.call(null, { res, text }),
-      onEventMessage: (eventMessage) => {
-        let data = null;
-        try {
-          data = JSON.parse(eventMessage.data);
-        } catch (e) {
-          return;
-        }
-
-        sendMessage(res, {
-          message: { ...userMessage, conversationId: data.conversation_id },
-          created: i === 0,
-        });
-
-        if (i === 0) {
-          i++;
-        }
-      },
-    });
-
-    logger.debug('[/ask/chatGPTBrowser]', response);
-
-    const newConversationId = response.conversationId || conversationId;
-    const newUserMassageId = response.parentMessageId || userMessageId;
-    const newResponseMessageId = response.messageId;
-
-    // STEP1 generate response message
-    response.text = response.response || '**ChatGPT refused to answer.**';
-
-    let responseMessage = {
-      conversationId: newConversationId,
-      messageId: responseMessageId,
-      newMessageId: newResponseMessageId,
-      parentMessageId: overrideParentMessageId || newUserMassageId,
-      text: await handleText(response),
-      sender: endpointOption?.chatGptLabel || 'ChatGPT',
-      unfinished: false,
-      error: false,
-      isCreatedByUser: false,
-    };
-
-    await saveMessage(req, { ...responseMessage, user });
-    responseMessage.messageId = newResponseMessageId;
-
-    // STEP2 update the conversation
-
-    // First update conversationId if needed
-    let conversationUpdate = { conversationId: newConversationId, endpoint: 'chatGPTBrowser' };
-    if (conversationId != newConversationId) {
-      if (isNewConversation) {
-        // change the conversationId to new one
-        conversationUpdate = {
-          ...conversationUpdate,
-          conversationId: conversationId,
-          newConversationId: newConversationId,
-        };
-      } else {
-        // create new conversation
-        conversationUpdate = {
-          ...conversationUpdate,
-          ...endpointOption,
-        };
-      }
-    }
-
-    await saveConvo(req, conversationUpdate);
-    conversationId = newConversationId;
-
-    // STEP3 update the user message
-    userMessage.conversationId = newConversationId;
-    userMessage.messageId = newUserMassageId;
-
-    // If response has parentMessageId, the fake userMessage.messageId should be updated to the real one.
-    if (!overrideParentMessageId) {
-      await saveMessage(req, {
-        ...userMessage,
-        user,
-        messageId: userMessageId,
-        newMessageId: newUserMassageId,
-      });
-    }
-    userMessageId = newUserMassageId;
-
-    sendMessage(res, {
-      title: await getConvoTitle(user, conversationId),
-      final: true,
-      conversation: await getConvo(user, conversationId),
-      requestMessage: userMessage,
-      responseMessage: responseMessage,
-    });
-    res.end();
-
-    if (userParentMessageId == Constants.NO_PARENT) {
-      // const title = await titleConvo({ endpoint: endpointOption?.endpoint, text, response: responseMessage });
-      const title = await response.details.title;
-      await saveConvo(req, {
-        conversationId: conversationId,
-        title,
-      });
-    }
-  } catch (error) {
-    const errorMessage = {
-      messageId: responseMessageId,
-      sender: 'ChatGPT',
-      conversationId,
-      parentMessageId: overrideParentMessageId || userMessageId,
-      unfinished: false,
-      error: true,
-      isCreatedByUser: false,
-      text: `${getPartialMessage() ?? ''}\n\nError message: "${error.message}"`,
-    };
-    await saveMessage(req, { ...errorMessage, user });
-    handleError(res, errorMessage);
-  }
-};
-
-module.exports = router;

api/server/routes/ask/bingAI.js

@@ -1,297 +0,0 @@
-const crypto = require('crypto');
-const express = require('express');
-const { Constants } = require('librechat-data-provider');
-const { handleError, sendMessage, createOnProgress, handleText } = require('~/server/utils');
-const { saveMessage, getConvoTitle, saveConvo, getConvo } = require('~/models');
-const { setHeaders } = require('~/server/middleware');
-const { titleConvoBing, askBing } = require('~/app');
-const { logger } = require('~/config');
-
-const router = express.Router();
-
-router.post('/', setHeaders, async (req, res) => {
-  const {
-    endpoint,
-    text,
-    messageId,
-    overrideParentMessageId = null,
-    parentMessageId,
-    conversationId: oldConversationId,
-  } = req.body;
-  if (text.length === 0) {
-    return handleError(res, { text: 'Prompt empty or too short' });
-  }
-  if (endpoint !== 'bingAI') {
-    return handleError(res, { text: 'Illegal request' });
-  }
-
-  // build user message
-  const conversationId = oldConversationId || crypto.randomUUID();
-  const isNewConversation = !oldConversationId;
-  const userMessageId = messageId;
-  const userParentMessageId = parentMessageId || Constants.NO_PARENT;
-  let userMessage = {
-    messageId: userMessageId,
-    sender: 'User',
-    text,
-    parentMessageId: userParentMessageId,
-    conversationId,
-    isCreatedByUser: true,
-  };
-
-  // build endpoint option
-  let endpointOption = {};
-  if (req.body?.jailbreak) {
-    endpointOption = {
-      jailbreak: req.body?.jailbreak ?? false,
-      jailbreakConversationId: req.body?.jailbreakConversationId ?? null,
-      systemMessage: req.body?.systemMessage ?? null,
-      context: req.body?.context ?? null,
-      toneStyle: req.body?.toneStyle ?? 'creative',
-      key: req.body?.key ?? null,
-    };
-  } else {
-    endpointOption = {
-      jailbreak: req.body?.jailbreak ?? false,
-      systemMessage: req.body?.systemMessage ?? null,
-      context: req.body?.context ?? null,
-      conversationSignature: req.body?.conversationSignature ?? null,
-      clientId: req.body?.clientId ?? null,
-      invocationId: req.body?.invocationId ?? null,
-      toneStyle: req.body?.toneStyle ?? 'creative',
-      key: req.body?.key ?? null,
-    };
-  }
-
-  logger.debug('[/ask/bingAI] ask log', {
-    userMessage,
-    endpointOption,
-    conversationId,
-  });
-
-  if (!overrideParentMessageId) {
-    await saveMessage(req, { ...userMessage, user: req.user.id });
-    await saveConvo(req, {
-      ...userMessage,
-      ...endpointOption,
-      conversationId,
-      endpoint,
-    });
-  }
-
-  // eslint-disable-next-line no-use-before-define
-  return await ask({
-    isNewConversation,
-    userMessage,
-    endpointOption,
-    conversationId,
-    preSendRequest: true,
-    overrideParentMessageId,
-    req,
-    res,
-  });
-});
-
-const ask = async ({
-  isNewConversation,
-  userMessage,
-  endpointOption,
-  conversationId,
-  preSendRequest = true,
-  overrideParentMessageId = null,
-  req,
-  res,
-}) => {
-  let { text, parentMessageId: userParentMessageId, messageId: userMessageId } = userMessage;
-  const user = req.user.id;
-
-  let responseMessageId = crypto.randomUUID();
-  const model = endpointOption?.jailbreak ? 'Sydney' : 'BingAI';
-
-  if (preSendRequest) {
-    sendMessage(res, { message: userMessage, created: true });
-  }
-
-  let lastSavedTimestamp = 0;
-  const { onProgress: progressCallback, getPartialText } = createOnProgress({
-    onProgress: ({ text }) => {
-      const currentTimestamp = Date.now();
-      if (currentTimestamp - lastSavedTimestamp > 500) {
-        lastSavedTimestamp = currentTimestamp;
-        saveMessage(req, {
-          messageId: responseMessageId,
-          sender: model,
-          conversationId,
-          parentMessageId: overrideParentMessageId || userMessageId,
-          model,
-          text: text,
-          unfinished: true,
-          error: false,
-          isCreatedByUser: false,
-          user,
-        });
-      }
-    },
-  });
-  const abortController = new AbortController();
-  let bingConversationId = null;
-  if (!isNewConversation) {
-    const convo = await getConvo(user, conversationId);
-    bingConversationId = convo.bingConversationId;
-  }
-
-  try {
-    let response = await askBing({
-      text,
-      userId: user,
-      parentMessageId: userParentMessageId,
-      conversationId: bingConversationId ?? conversationId,
-      ...endpointOption,
-      onProgress: progressCallback.call(null, {
-        res,
-        text,
-        parentMessageId: overrideParentMessageId || userMessageId,
-      }),
-      abortController,
-    });
-
-    logger.debug('[/ask/bingAI] BING RESPONSE', response);
-
-    if (response.details && response.details.scores) {
-      logger.debug('[/ask/bingAI] SCORES', response.details.scores);
-    }
-
-    const newConversationId = endpointOption?.jailbreak
-      ? response.jailbreakConversationId
-      : response.conversationId || conversationId;
-    const newUserMessageId =
-      response.parentMessageId || response.details.requestId || userMessageId;
-    const newResponseMessageId = response.messageId || response.details.messageId;
-
-    // STEP1 generate response message
-    response.text =
-      response.response || response.details.spokenText || '**Bing refused to answer.**';
-
-    const partialText = getPartialText();
-    let unfinished = false;
-    if (partialText?.trim()?.length > response.text.length) {
-      response.text = partialText;
-      unfinished = false;
-      //setting "unfinished" to false fix bing image generation error msg and allows to continue a convo after being triggered by censorship (bing does remember the context after a "censored error" so there is no reason to end the convo)
-    }
-
-    let responseMessage = {
-      conversationId,
-      bingConversationId: newConversationId,
-      messageId: responseMessageId,
-      newMessageId: newResponseMessageId,
-      parentMessageId: overrideParentMessageId || newUserMessageId,
-      sender: model,
-      text: await handleText(response, true),
-      model,
-      suggestions:
-        response.details.suggestedResponses &&
-        response.details.suggestedResponses.map((s) => s.text),
-      unfinished,
-      error: false,
-      isCreatedByUser: false,
-    };
-
-    await saveMessage(req, { ...responseMessage, user });
-    responseMessage.messageId = newResponseMessageId;
-
-    let conversationUpdate = {
-      conversationId,
-      bingConversationId: newConversationId,
-      endpoint: 'bingAI',
-    };
-
-    if (endpointOption?.jailbreak) {
-      conversationUpdate.jailbreak = true;
-      conversationUpdate.jailbreakConversationId = response.jailbreakConversationId;
-    } else {
-      conversationUpdate.jailbreak = false;
-      conversationUpdate.conversationSignature = response.encryptedConversationSignature;
-      conversationUpdate.clientId = response.clientId;
-      conversationUpdate.invocationId = response.invocationId;
-    }
-
-    await saveConvo(req, conversationUpdate);
-    userMessage.messageId = newUserMessageId;
-
-    // If response has parentMessageId, the fake userMessage.messageId should be updated to the real one.
-    if (!overrideParentMessageId) {
-      await saveMessage(req, {
-        ...userMessage,
-        user,
-        messageId: userMessageId,
-        newMessageId: newUserMessageId,
-      });
-    }
-    userMessageId = newUserMessageId;
-
-    sendMessage(res, {
-      title: await getConvoTitle(user, conversationId),
-      final: true,
-      conversation: await getConvo(user, conversationId),
-      requestMessage: userMessage,
-      responseMessage: responseMessage,
-    });
-    res.end();
-
-    if (userParentMessageId == Constants.NO_PARENT) {
-      const title = await titleConvoBing({
-        text,
-        response: responseMessage,
-      });
-
-      await saveConvo(req, {
-        conversationId: conversationId,
-        title,
-      });
-    }
-  } catch (error) {
-    logger.error('[/ask/bingAI] Error handling BingAI response', error);
-    const partialText = getPartialText();
-    if (partialText?.length > 2) {
-      const responseMessage = {
-        messageId: responseMessageId,
-        sender: model,
-        conversationId,
-        parentMessageId: overrideParentMessageId || userMessageId,
-        text: partialText,
-        model,
-        unfinished: true,
-        error: false,
-        isCreatedByUser: false,
-      };
-
-      saveMessage(req, { ...responseMessage, user });
-
-      return {
-        title: await getConvoTitle(user, conversationId),
-        final: true,
-        conversation: await getConvo(user, conversationId),
-        requestMessage: userMessage,
-        responseMessage: responseMessage,
-      };
-    } else {
-      logger.error('[/ask/bingAI] Error handling BingAI response', error);
-      const errorMessage = {
-        messageId: responseMessageId,
-        sender: model,
-        conversationId,
-        parentMessageId: overrideParentMessageId || userMessageId,
-        unfinished: false,
-        error: true,
-        text: error.message,
-        model,
-        isCreatedByUser: false,
-      };
-      await saveMessage(req, { ...errorMessage, user });
-      handleError(res, errorMessage);
-    }
-  }
-};
-
-module.exports = router;

api/server/routes/ask/gptPlugins.js

@@ -1,11 +1,9 @@
 const express = require('express');
-const throttle = require('lodash/throttle');
-const { getResponseSender, Constants, CacheKeys, Time } = require('librechat-data-provider');
+const { getResponseSender, Constants } = require('librechat-data-provider');
 const { initializeClient } = require('~/server/services/Endpoints/gptPlugins');
 const { sendMessage, createOnProgress } = require('~/server/utils');
 const { addTitle } = require('~/server/services/Endpoints/openAI');
 const { saveMessage, updateMessage } = require('~/models');
-const { getLogStores } = require('~/cache');
 const {
   handleAbort,
   createAbortController,
@@ -72,15 +70,6 @@ router.post(
       }
     };
 
-    const messageCache = getLogStores(CacheKeys.MESSAGES);
-    const throttledCacheSet = throttle(
-      (text) => {
-        messageCache.set(responseMessageId, text, Time.FIVE_MINUTES);
-      },
-      3000,
-      { trailing: false },
-    );
-
     let streaming = null;
     let timer = null;
 
@@ -89,13 +78,11 @@ router.post(
       sendIntermediateMessage,
       getPartialText,
     } = createOnProgress({
-      onProgress: ({ text: partialText }) => {
+      onProgress: () => {
         if (timer) {
           clearTimeout(timer);
         }
 
-        throttledCacheSet(partialText);
-
         streaming = new Promise((resolve) => {
           timer = setTimeout(() => {
             resolve();

api/server/routes/ask/index.js

@@ -2,10 +2,8 @@ const express = require('express');
 const openAI = require('./openAI');
 const custom = require('./custom');
 const google = require('./google');
-const bingAI = require('./bingAI');
 const anthropic = require('./anthropic');
 const gptPlugins = require('./gptPlugins');
-const askChatGPTBrowser = require('./askChatGPTBrowser');
 const { isEnabled } = require('~/server/utils');
 const { EModelEndpoint } = require('librechat-data-provider');
 const {
@@ -41,11 +39,9 @@ if (isEnabled(LIMIT_MESSAGE_USER)) {
 router.use(validateConvoAccess);
 
 router.use([`/${EModelEndpoint.azureOpenAI}`, `/${EModelEndpoint.openAI}`], openAI);
-router.use(`/${EModelEndpoint.chatGPTBrowser}`, askChatGPTBrowser);
 router.use(`/${EModelEndpoint.gptPlugins}`, gptPlugins);
 router.use(`/${EModelEndpoint.anthropic}`, anthropic);
 router.use(`/${EModelEndpoint.google}`, google);
-router.use(`/${EModelEndpoint.bingAI}`, bingAI);
 router.use(`/${EModelEndpoint.custom}`, custom);
 
 module.exports = router;

api/server/routes/assistants/actions.js

@@ -1,6 +1,6 @@
 const express = require('express');
 const { nanoid } = require('nanoid');
-const { actionDelimiter, EModelEndpoint } = require('librechat-data-provider');
+const { actionDelimiter, EModelEndpoint, removeNullishValues } = require('librechat-data-provider');
 const { encryptMetadata, domainParser } = require('~/server/services/ActionService');
 const { getOpenAIClient } = require('~/server/controllers/assistants/helpers');
 const { updateAction, getActions, deleteAction } = require('~/models/Action');
@@ -29,7 +29,7 @@ router.post('/:assistant_id', async (req, res) => {
       return res.status(400).json({ message: 'No functions provided' });
     }
 
-    let metadata = await encryptMetadata(_metadata);
+    let metadata = await encryptMetadata(removeNullishValues(_metadata, true));
     const isDomainAllowed = await isActionDomainAllowed(metadata.domain);
     if (!isDomainAllowed) {
       return res.status(400).json({ message: 'Domain not allowed' });
@@ -63,7 +63,7 @@ router.post('/:assistant_id', async (req, res) => {
       return res.status(404).json({ message: 'Assistant not found' });
     }
 
-    const { actions: _actions = [] } = assistant_data ?? {};
+    const { actions: _actions = [], user: assistant_user } = assistant_data ?? {};
     const actions = [];
     for (const action of _actions) {
       const [_action_domain, current_action_id] = action.split(actionDelimiter);
@@ -99,16 +99,21 @@ router.post('/:assistant_id', async (req, res) => {
 
     let updatedAssistant = await openai.beta.assistants.update(assistant_id, { tools });
     const promises = [];
-    promises.push(
-      updateAssistantDoc(
-        { assistant_id },
-        {
-          actions,
-          user: req.user.id,
-        },
-      ),
-    );
-    promises.push(updateAction({ action_id }, { metadata, assistant_id, user: req.user.id }));
+
+    // Only update user field for new assistant documents
+    const assistantUpdateData = { actions };
+    if (!assistant_data) {
+      assistantUpdateData.user = req.user.id;
+    }
+    promises.push(updateAssistantDoc({ assistant_id }, assistantUpdateData));
+
+    // Only update user field for new actions
+    const actionUpdateData = { metadata, assistant_id };
+    if (!actions_result || !actions_result.length) {
+      // For new actions, use the assistant owner's user ID
+      actionUpdateData.user = assistant_user || req.user.id;
+    }
+    promises.push(updateAction({ action_id }, actionUpdateData));
 
     /** @type {[AssistantDocument, Action]} */
     let [assistantDocument, updatedAction] = await Promise.all(promises);
@@ -180,15 +185,12 @@ router.delete('/:assistant_id/:action_id/:model', async (req, res) => {
     await openai.beta.assistants.update(assistant_id, { tools: updatedTools });
 
     const promises = [];
-    promises.push(
-      updateAssistantDoc(
-        { assistant_id },
-        {
-          actions: updatedActions,
-          user: req.user.id,
-        },
-      ),
-    );
+    // Only update user field if assistant document doesn't exist
+    const assistantUpdateData = { actions: updatedActions };
+    if (!assistant_data) {
+      assistantUpdateData.user = req.user.id;
+    }
+    promises.push(updateAssistantDoc({ assistant_id }, assistantUpdateData));
     promises.push(deleteAction({ action_id }));
 
     await Promise.all(promises);

api/server/routes/config.js

@@ -46,6 +46,11 @@ router.get('/', async function (req, res) {
         !!process.env.FACEBOOK_CLIENT_ID && !!process.env.FACEBOOK_CLIENT_SECRET,
       githubLoginEnabled: !!process.env.GITHUB_CLIENT_ID && !!process.env.GITHUB_CLIENT_SECRET,
       googleLoginEnabled: !!process.env.GOOGLE_CLIENT_ID && !!process.env.GOOGLE_CLIENT_SECRET,
+      appleLoginEnabled:
+          !!process.env.APPLE_CLIENT_ID &&
+          !!process.env.APPLE_TEAM_ID &&
+          !!process.env.APPLE_KEY_ID &&
+          !!process.env.APPLE_PRIVATE_KEY_PATH,
       openidLoginEnabled:
         !!process.env.OPENID_CLIENT_ID &&
         !!process.env.OPENID_CLIENT_SECRET &&

api/server/routes/edit/gptPlugins.js

@@ -1,6 +1,5 @@
 const express = require('express');
-const throttle = require('lodash/throttle');
-const { getResponseSender, CacheKeys, Time } = require('librechat-data-provider');
+const { getResponseSender } = require('librechat-data-provider');
 const {
   setHeaders,
   handleAbort,
@@ -14,7 +13,6 @@ const {
 const { sendMessage, createOnProgress, formatSteps, formatAction } = require('~/server/utils');
 const { initializeClient } = require('~/server/services/Endpoints/gptPlugins');
 const { saveMessage, updateMessage } = require('~/models');
-const { getLogStores } = require('~/cache');
 const { validateTools } = require('~/app');
 const { logger } = require('~/config');
 
@@ -80,26 +78,16 @@ router.post(
       }
     };
 
-    const messageCache = getLogStores(CacheKeys.MESSAGES);
-    const throttledCacheSet = throttle(
-      (text) => {
-        messageCache.set(responseMessageId, text, Time.FIVE_MINUTES);
-      },
-      3000,
-      { trailing: false },
-    );
-
     const {
       onProgress: progressCallback,
       sendIntermediateMessage,
       getPartialText,
     } = createOnProgress({
       generation,
-      onProgress: ({ text: partialText }) => {
+      onProgress: () => {
         if (plugin.loading === true) {
           plugin.loading = false;
         }
-        throttledCacheSet(partialText);
       },
     });
 

api/server/routes/files/speech/customConfigSpeech.js

@@ -3,7 +3,7 @@ const router = express.Router();
 
 const { getCustomConfigSpeech } = require('~/server/services/Files/Audio');
 
-router.get('/get', async (req, res) => {
+router.get('/', async (req, res) => {
   await getCustomConfigSpeech(req, res);
 });
 

api/server/routes/files/speech/index.js

@@ -4,6 +4,7 @@ const { createTTSLimiters, createSTTLimiters } = require('~/server/middleware');
 const stt = require('./stt');
 const tts = require('./tts');
 const customConfigSpeech = require('./customConfigSpeech');
+const realtime = require('./realtime');
 
 const router = express.Router();
 
@@ -14,4 +15,6 @@ router.use('/tts', ttsIpLimiter, ttsUserLimiter, tts);
 
 router.use('/config', customConfigSpeech);
 
+router.use('/realtime', realtime);
+
 module.exports = router;

api/server/routes/files/speech/realtime.js

@@ -0,0 +1,10 @@
+const express = require('express');
+const router = express.Router();
+
+const { getRealtimeConfig } = require('~/server/services/Files/Audio');
+
+router.get('/', async (req, res) => {
+  await getRealtimeConfig(req, res);
+});
+
+module.exports = router;

api/server/routes/index.js

@@ -2,6 +2,7 @@ const assistants = require('./assistants');
 const categories = require('./categories');
 const tokenizer = require('./tokenizer');
 const endpoints = require('./endpoints');
+const websocket = require('./websocket');
 const staticRoute = require('./static');
 const messages = require('./messages');
 const presets = require('./presets');
@@ -9,11 +10,13 @@ const prompts = require('./prompts');
 const balance = require('./balance');
 const plugins = require('./plugins');
 const bedrock = require('./bedrock');
+const actions = require('./actions');
 const search = require('./search');
 const models = require('./models');
 const convos = require('./convos');
 const config = require('./config');
 const agents = require('./agents');
+const banner = require('./banner');
 const roles = require('./roles');
 const oauth = require('./oauth');
 const files = require('./files');
@@ -24,7 +27,6 @@ const edit = require('./edit');
 const keys = require('./keys');
 const user = require('./user');
 const ask = require('./ask');
-const banner = require('./banner');
 
 module.exports = {
   ask,
@@ -38,6 +40,7 @@ module.exports = {
   files,
   share,
   agents,
+  banner,
   bedrock,
   convos,
   search,
@@ -45,13 +48,14 @@ module.exports = {
   config,
   models,
   plugins,
+  actions,
   presets,
   balance,
   messages,
+  websocket,
   endpoints,
   tokenizer,
   assistants,
   categories,
   staticRoute,
-  banner,
 };

api/server/routes/messages.js

@@ -1,6 +1,14 @@
 const express = require('express');
 const { ContentTypes } = require('librechat-data-provider');
-const { saveConvo, saveMessage, getMessages, updateMessage, deleteMessages } = require('~/models');
+const {
+  saveConvo,
+  saveMessage,
+  getMessage,
+  getMessages,
+  updateMessage,
+  deleteMessages,
+} = require('~/models');
+const { findAllArtifacts, replaceArtifactContent } = require('~/server/services/Artifacts/update');
 const { requireJwtAuth, validateMessageReq } = require('~/server/middleware');
 const { countTokens } = require('~/server/utils');
 const { logger } = require('~/config');
@@ -8,6 +16,68 @@ const { logger } = require('~/config');
 const router = express.Router();
 router.use(requireJwtAuth);
 
+router.post('/artifact/:messageId', async (req, res) => {
+  try {
+    const { messageId } = req.params;
+    const { index, original, updated } = req.body;
+
+    if (typeof index !== 'number' || index < 0 || original == null || updated == null) {
+      return res.status(400).json({ error: 'Invalid request parameters' });
+    }
+
+    const message = await getMessage({ user: req.user.id, messageId });
+    if (!message) {
+      return res.status(404).json({ error: 'Message not found' });
+    }
+
+    const artifacts = findAllArtifacts(message);
+    if (index >= artifacts.length) {
+      return res.status(400).json({ error: 'Artifact index out of bounds' });
+    }
+
+    const targetArtifact = artifacts[index];
+    let updatedText = null;
+
+    if (targetArtifact.source === 'content') {
+      const part = message.content[targetArtifact.partIndex];
+      updatedText = replaceArtifactContent(part.text, targetArtifact, original, updated);
+      if (updatedText) {
+        part.text = updatedText;
+      }
+    } else {
+      updatedText = replaceArtifactContent(message.text, targetArtifact, original, updated);
+      if (updatedText) {
+        message.text = updatedText;
+      }
+    }
+
+    if (!updatedText) {
+      return res.status(400).json({ error: 'Original content not found in target artifact' });
+    }
+
+    const savedMessage = await saveMessage(
+      req,
+      {
+        messageId,
+        conversationId: message.conversationId,
+        text: message.text,
+        content: message.content,
+        user: req.user.id,
+      },
+      { context: 'POST /api/messages/artifact/:messageId' },
+    );
+
+    res.status(200).json({
+      conversationId: savedMessage.conversationId,
+      content: savedMessage.content,
+      text: savedMessage.text,
+    });
+  } catch (error) {
+    logger.error('Error editing artifact:', error);
+    res.status(500).json({ error: 'Internal server error' });
+  }
+});
+
 /* Note: It's necessary to add `validateMessageReq` within route definition for correct params */
 router.get('/:conversationId', validateMessageReq, async (req, res) => {
   try {

api/server/routes/oauth.js

@@ -56,6 +56,9 @@ router.get(
   oauthHandler,
 );
 
+/**
+ * Facebook Routes
+ */
 router.get(
   '/facebook',
   passport.authenticate('facebook', {
@@ -77,6 +80,9 @@ router.get(
   oauthHandler,
 );
 
+/**
+ * OpenID Routes
+ */
 router.get(
   '/openid',
   passport.authenticate('openid', {
@@ -94,6 +100,9 @@ router.get(
   oauthHandler,
 );
 
+/**
+ * GitHub Routes
+ */
 router.get(
   '/github',
   passport.authenticate('github', {
@@ -112,6 +121,10 @@ router.get(
   }),
   oauthHandler,
 );
+
+/**
+ * Discord Routes
+ */
 router.get(
   '/discord',
   passport.authenticate('discord', {
@@ -131,4 +144,24 @@ router.get(
   oauthHandler,
 );
 
+/**
+ * Apple Routes
+ */
+router.get(
+  '/apple',
+  passport.authenticate('apple', {
+    session: false,
+  }),
+);
+
+router.post(
+  '/apple/callback',
+  passport.authenticate('apple', {
+    failureRedirect: `${domains.client}/oauth/error`,
+    failureMessage: true,
+    session: false,
+  }),
+  oauthHandler,
+);
+
 module.exports = router;

api/server/routes/prompts.js

@@ -214,10 +214,7 @@ const deletePromptController = async (req, res) => {
     const { promptId } = req.params;
     const { groupId } = req.query;
     const author = req.user.id;
-    const query = { promptId, groupId, author };
-    if (req.user.role === SystemRoles.ADMIN) {
-      delete query.author;
-    }
+    const query = { promptId, groupId, author, role: req.user.role };
     const result = await deletePrompt(query);
     res.status(200).send(result);
   } catch (error) {

api/server/routes/share.js

@@ -1,6 +1,7 @@
 const express = require('express');
 
 const {
+  getSharedLink,
   getSharedMessages,
   createSharedLink,
   updateSharedLink,
@@ -45,29 +46,60 @@ if (allowSharedLinks) {
  */
 router.get('/', requireJwtAuth, async (req, res) => {
   try {
-    let pageNumber = req.query.pageNumber || 1;
-    pageNumber = parseInt(pageNumber, 10);
+    const params = {
+      pageParam: req.query.cursor,
+      pageSize: Math.max(1, parseInt(req.query.pageSize) || 10),
+      isPublic: isEnabled(req.query.isPublic),
+      sortBy: ['createdAt', 'title'].includes(req.query.sortBy) ? req.query.sortBy : 'createdAt',
+      sortDirection: ['asc', 'desc'].includes(req.query.sortDirection)
+        ? req.query.sortDirection
+        : 'desc',
+      search: req.query.search
+        ? decodeURIComponent(req.query.search.trim())
+        : undefined,
+    };
 
-    if (isNaN(pageNumber) || pageNumber < 1) {
-      return res.status(400).json({ error: 'Invalid page number' });
-    }
+    const result = await getSharedLinks(
+      req.user.id,
+      params.pageParam,
+      params.pageSize,
+      params.isPublic,
+      params.sortBy,
+      params.sortDirection,
+      params.search,
+    );
 
-    let pageSize = req.query.pageSize || 25;
-    pageSize = parseInt(pageSize, 10);
-
-    if (isNaN(pageSize) || pageSize < 1) {
-      return res.status(400).json({ error: 'Invalid page size' });
-    }
-    const isPublic = req.query.isPublic === 'true';
-    res.status(200).send(await getSharedLinks(req.user.id, pageNumber, pageSize, isPublic));
+    res.status(200).send({
+      links: result.links,
+      nextCursor: result.nextCursor,
+      hasNextPage: result.hasNextPage,
+    });
   } catch (error) {
-    res.status(500).json({ message: 'Error getting shared links' });
+    console.error('Error getting shared links:', error);
+    res.status(500).json({
+      message: 'Error getting shared links',
+      error: error.message,
+    });
   }
 });
 
-router.post('/', requireJwtAuth, async (req, res) => {
+router.get('/link/:conversationId', requireJwtAuth, async (req, res) => {
   try {
-    const created = await createSharedLink(req.user.id, req.body);
+    const share = await getSharedLink(req.user.id, req.params.conversationId);
+
+    return res.status(200).json({
+      success: share.success,
+      shareId: share.shareId,
+      conversationId: req.params.conversationId,
+    });
+  } catch (error) {
+    res.status(500).json({ message: 'Error getting shared link' });
+  }
+});
+
+router.post('/:conversationId', requireJwtAuth, async (req, res) => {
+  try {
+    const created = await createSharedLink(req.user.id, req.params.conversationId);
     if (created) {
       res.status(200).json(created);
     } else {
@@ -78,11 +110,11 @@ router.post('/', requireJwtAuth, async (req, res) => {
   }
 });
 
-router.patch('/', requireJwtAuth, async (req, res) => {
+router.patch('/:shareId', requireJwtAuth, async (req, res) => {
   try {
-    const updated = await updateSharedLink(req.user.id, req.body);
-    if (updated) {
-      res.status(200).json(updated);
+    const updatedShare = await updateSharedLink(req.user.id, req.params.shareId);
+    if (updatedShare) {
+      res.status(200).json(updatedShare);
     } else {
       res.status(404).end();
     }
@@ -93,14 +125,15 @@ router.patch('/', requireJwtAuth, async (req, res) => {
 
 router.delete('/:shareId', requireJwtAuth, async (req, res) => {
   try {
-    const deleted = await deleteSharedLink(req.user.id, { shareId: req.params.shareId });
-    if (deleted) {
-      res.status(200).json(deleted);
-    } else {
-      res.status(404).end();
+    const result = await deleteSharedLink(req.user.id, req.params.shareId);
+
+    if (!result) {
+      return res.status(404).json({ message: 'Share not found' });
     }
+
+    return res.status(200).json(result);
   } catch (error) {
-    res.status(500).json({ message: 'Error deleting shared link' });
+    return res.status(400).json({ message: error.message });
   }
 });
 

api/server/routes/websocket.js

@@ -0,0 +1,19 @@
+const express = require('express');
+const optionalJwtAuth = require('~/server/middleware/optionalJwtAuth');
+const router = express.Router();
+
+router.get('/', optionalJwtAuth, async (req, res) => {
+  const isProduction = process.env.NODE_ENV === 'production';
+
+  const protocol = isProduction && req.secure ? 'https' : 'http';
+
+  const serverDomain = process.env.SERVER_DOMAIN
+    ? process.env.SERVER_DOMAIN.replace(/^https?:\/\//, '')
+    : req.headers.host;
+
+  const socketIoUrl = `${protocol}://${serverDomain}`;
+
+  res.json({ url: socketIoUrl });
+});
+
+module.exports = router;

api/server/services/ActionService.js

@@ -1,20 +1,28 @@
+const jwt = require('jsonwebtoken');
+const { nanoid } = require('nanoid');
+const { tool } = require('@langchain/core/tools');
+const { GraphEvents, sleep } = require('@librechat/agents');
 const {
+  Time,
   CacheKeys,
+  StepTypes,
   Constants,
   AuthTypeEnum,
   actionDelimiter,
   isImageVisionTool,
   actionDomainSeparator,
 } = require('librechat-data-provider');
-const { tool } = require('@langchain/core/tools');
+const { refreshAccessToken } = require('~/server/services/TokenService');
 const { isActionDomainAllowed } = require('~/server/services/domains');
+const { logger, getFlowStateManager, sendEvent } = require('~/config');
 const { encryptV2, decryptV2 } = require('~/server/utils/crypto');
 const { getActions, deleteActions } = require('~/models/Action');
 const { deleteAssistant } = require('~/models/Assistant');
+const { findToken } = require('~/models/Token');
 const { logAxiosError } = require('~/utils');
 const { getLogStores } = require('~/cache');
-const { logger } = require('~/config');
 
+const JWT_SECRET = process.env.JWT_SECRET;
 const toolNameRegex = /^[a-zA-Z0-9_-]+$/;
 const replaceSeparatorRegex = new RegExp(actionDomainSeparator, 'g');
 
@@ -115,6 +123,8 @@ async function loadActionSets(searchParams) {
  * Creates a general tool for an entire action set.
  *
  * @param {Object} params - The parameters for loading action sets.
+ * @param {ServerRequest} params.req
+ * @param {ServerResponse} params.res
  * @param {Action} params.action - The action set. Necessary for decrypting authentication values.
  * @param {ActionRequest} params.requestBuilder - The ActionRequest builder class to execute the API call.
  * @param {string | undefined} [params.name] - The name of the tool.
@@ -122,33 +132,185 @@ async function loadActionSets(searchParams) {
  * @param {import('zod').ZodTypeAny | undefined} [params.zodSchema] - The Zod schema for tool input validation/definition
  * @returns { Promise<typeof tool | { _call: (toolInput: Object | string) => unknown}> } An object with `_call` method to execute the tool input.
  */
-async function createActionTool({ action, requestBuilder, zodSchema, name, description }) {
-  action.metadata = await decryptMetadata(action.metadata);
+async function createActionTool({
+  req,
+  res,
+  action,
+  requestBuilder,
+  zodSchema,
+  name,
+  description,
+}) {
   const isDomainAllowed = await isActionDomainAllowed(action.metadata.domain);
   if (!isDomainAllowed) {
     return null;
   }
-  /** @type {(toolInput: Object | string) => Promise<unknown>} */
-  const _call = async (toolInput) => {
-    try {
-      const executor = requestBuilder.createExecutor();
+  const encrypted = {
+    oauth_client_id: action.metadata.oauth_client_id,
+    oauth_client_secret: action.metadata.oauth_client_secret,
+  };
+  action.metadata = await decryptMetadata(action.metadata);
 
-      // Chain the operations
+  /** @type {(toolInput: Object | string, config: GraphRunnableConfig) => Promise<unknown>} */
+  const _call = async (toolInput, config) => {
+    try {
+      /** @type {import('librechat-data-provider').ActionMetadataRuntime} */
+      const metadata = action.metadata;
+      const executor = requestBuilder.createExecutor();
       const preparedExecutor = executor.setParams(toolInput);
 
-      if (action.metadata.auth && action.metadata.auth.type !== AuthTypeEnum.None) {
-        await preparedExecutor.setAuth(action.metadata);
+      if (metadata.auth && metadata.auth.type !== AuthTypeEnum.None) {
+        try {
+          const action_id = action.action_id;
+          const identifier = `${req.user.id}:${action.action_id}`;
+          if (metadata.auth.type === AuthTypeEnum.OAuth && metadata.auth.authorization_url) {
+            const requestLogin = async () => {
+              const { args: _args, stepId, ...toolCall } = config.toolCall ?? {};
+              if (!stepId) {
+                throw new Error('Tool call is missing stepId');
+              }
+              const statePayload = {
+                nonce: nanoid(),
+                user: req.user.id,
+                action_id,
+              };
+
+              const stateToken = jwt.sign(statePayload, JWT_SECRET, { expiresIn: '10m' });
+              try {
+                const redirectUri = `${process.env.DOMAIN_CLIENT}/api/actions/${action_id}/oauth/callback`;
+                const params = new URLSearchParams({
+                  client_id: metadata.oauth_client_id,
+                  scope: metadata.auth.scope,
+                  redirect_uri: redirectUri,
+                  access_type: 'offline',
+                  response_type: 'code',
+                  state: stateToken,
+                });
+
+                const authURL = `${metadata.auth.authorization_url}?${params.toString()}`;
+                /** @type {{ id: string; delta: AgentToolCallDelta }} */
+                const data = {
+                  id: stepId,
+                  delta: {
+                    type: StepTypes.TOOL_CALLS,
+                    tool_calls: [{ ...toolCall, args: '' }],
+                    auth: authURL,
+                    expires_at: Date.now() + Time.TWO_MINUTES,
+                  },
+                };
+                const flowManager = await getFlowStateManager(getLogStores);
+                await flowManager.createFlowWithHandler(
+                  `${identifier}:login`,
+                  'oauth_login',
+                  async () => {
+                    sendEvent(res, { event: GraphEvents.ON_RUN_STEP_DELTA, data });
+                    logger.debug('Sent OAuth login request to client', { action_id, identifier });
+                    return true;
+                  },
+                );
+                logger.debug('Waiting for OAuth Authorization response', { action_id, identifier });
+                const result = await flowManager.createFlow(identifier, 'oauth', {
+                  state: stateToken,
+                  userId: req.user.id,
+                  client_url: metadata.auth.client_url,
+                  redirect_uri: `${process.env.DOMAIN_CLIENT}/api/actions/${action_id}/oauth/callback`,
+                  /** Encrypted values */
+                  encrypted_oauth_client_id: encrypted.oauth_client_id,
+                  encrypted_oauth_client_secret: encrypted.oauth_client_secret,
+                });
+                logger.debug('Received OAuth Authorization response', { action_id, identifier });
+                data.delta.auth = undefined;
+                data.delta.expires_at = undefined;
+                sendEvent(res, { event: GraphEvents.ON_RUN_STEP_DELTA, data });
+                await sleep(3000);
+                metadata.oauth_access_token = result.access_token;
+                metadata.oauth_refresh_token = result.refresh_token;
+                const expiresAt = new Date(Date.now() + result.expires_in * 1000);
+                metadata.oauth_token_expires_at = expiresAt.toISOString();
+              } catch (error) {
+                const errorMessage = 'Failed to authenticate OAuth tool';
+                logger.error(errorMessage, error);
+                throw new Error(errorMessage);
+              }
+            };
+
+            const tokenPromises = [];
+            tokenPromises.push(findToken({ userId: req.user.id, type: 'oauth', identifier }));
+            tokenPromises.push(
+              findToken({
+                userId: req.user.id,
+                type: 'oauth_refresh',
+                identifier: `${identifier}:refresh`,
+              }),
+            );
+            const [tokenData, refreshTokenData] = await Promise.all(tokenPromises);
+
+            if (tokenData) {
+              // Valid token exists, add it to metadata for setAuth
+              metadata.oauth_access_token = await decryptV2(tokenData.token);
+              if (refreshTokenData) {
+                metadata.oauth_refresh_token = await decryptV2(refreshTokenData.token);
+              }
+              metadata.oauth_token_expires_at = tokenData.expiresAt.toISOString();
+            } else if (!refreshTokenData) {
+              // No tokens exist, need to authenticate
+              await requestLogin();
+            } else if (refreshTokenData) {
+              // Refresh token is still valid, use it to get new access token
+              try {
+                const refresh_token = await decryptV2(refreshTokenData.token);
+                const refreshTokens = async () =>
+                  await refreshAccessToken({
+                    identifier,
+                    refresh_token,
+                    userId: req.user.id,
+                    client_url: metadata.auth.client_url,
+                    encrypted_oauth_client_id: encrypted.oauth_client_id,
+                    encrypted_oauth_client_secret: encrypted.oauth_client_secret,
+                  });
+                const flowManager = await getFlowStateManager(getLogStores);
+                const refreshData = await flowManager.createFlowWithHandler(
+                  `${identifier}:refresh`,
+                  'oauth_refresh',
+                  refreshTokens,
+                );
+                metadata.oauth_access_token = refreshData.access_token;
+                if (refreshData.refresh_token) {
+                  metadata.oauth_refresh_token = refreshData.refresh_token;
+                }
+                const expiresAt = new Date(Date.now() + refreshData.expires_in * 1000);
+                metadata.oauth_token_expires_at = expiresAt.toISOString();
+              } catch (error) {
+                logger.error('Failed to refresh token, requesting new login:', error);
+                await requestLogin();
+              }
+            } else {
+              await requestLogin();
+            }
+          }
+
+          await preparedExecutor.setAuth(metadata);
+        } catch (error) {
+          if (
+            error.message.includes('No access token found') ||
+            error.message.includes('Access token is expired')
+          ) {
+            throw error;
+          }
+          throw new Error(`Authentication failed: ${error.message}`);
+        }
       }
 
-      const res = await preparedExecutor.execute();
+      const response = await preparedExecutor.execute();
 
-      if (typeof res.data === 'object') {
-        return JSON.stringify(res.data);
+      if (typeof response.data === 'object') {
+        return JSON.stringify(response.data);
       }
-      return res.data;
+      return response.data;
     } catch (error) {
       const logMessage = `API call to ${action.metadata.domain} failed`;
       logAxiosError({ message: logMessage, error });
+      throw error;
     }
   };
 

api/server/services/Artifacts/update.js

@@ -0,0 +1,109 @@
+const ARTIFACT_START = ':::artifact';
+const ARTIFACT_END = ':::';
+
+/**
+ * Find all artifact boundaries in the message
+ * @param {TMessage} message
+ * @returns {Array<{start: number, end: number, source: 'content'|'text', partIndex?: number}>}
+ */
+const findAllArtifacts = (message) => {
+  const artifacts = [];
+
+  // Check content parts first
+  if (message.content?.length) {
+    message.content.forEach((part, partIndex) => {
+      if (part.type === 'text' && typeof part.text === 'string') {
+        let currentIndex = 0;
+        let start = part.text.indexOf(ARTIFACT_START, currentIndex);
+
+        while (start !== -1) {
+          const end = part.text.indexOf(ARTIFACT_END, start + ARTIFACT_START.length);
+          artifacts.push({
+            start,
+            end: end !== -1 ? end + ARTIFACT_END.length : part.text.length,
+            source: 'content',
+            partIndex,
+            text: part.text,
+          });
+
+          currentIndex = end !== -1 ? end + ARTIFACT_END.length : part.text.length;
+          start = part.text.indexOf(ARTIFACT_START, currentIndex);
+        }
+      }
+    });
+  }
+
+  // Check message.text if no content parts
+  if (!artifacts.length && message.text) {
+    let currentIndex = 0;
+    let start = message.text.indexOf(ARTIFACT_START, currentIndex);
+
+    while (start !== -1) {
+      const end = message.text.indexOf(ARTIFACT_END, start + ARTIFACT_START.length);
+      artifacts.push({
+        start,
+        end: end !== -1 ? end + ARTIFACT_END.length : message.text.length,
+        source: 'text',
+        text: message.text,
+      });
+
+      currentIndex = end !== -1 ? end + ARTIFACT_END.length : message.text.length;
+      start = message.text.indexOf(ARTIFACT_START, currentIndex);
+    }
+  }
+
+  return artifacts;
+};
+
+const replaceArtifactContent = (originalText, artifact, original, updated) => {
+  const artifactContent = artifact.text.substring(artifact.start, artifact.end);
+
+  // Find boundaries between ARTIFACT_START and ARTIFACT_END
+  const contentStart = artifactContent.indexOf('\n', artifactContent.indexOf(ARTIFACT_START)) + 1;
+  const contentEnd = artifactContent.lastIndexOf(ARTIFACT_END);
+
+  if (contentStart === -1 || contentEnd === -1) {
+    return null;
+  }
+
+  // Check if there are code blocks
+  const codeBlockStart = artifactContent.indexOf('```\n', contentStart);
+  const codeBlockEnd = artifactContent.lastIndexOf('\n```', contentEnd);
+
+  // Determine where to look for the original content
+  let searchStart, searchEnd;
+  if (codeBlockStart !== -1 && codeBlockEnd !== -1) {
+    // If code blocks exist, search between them
+    searchStart = codeBlockStart + 4; // after ```\n
+    searchEnd = codeBlockEnd;
+  } else {
+    // Otherwise search in the whole artifact content
+    searchStart = contentStart;
+    searchEnd = contentEnd;
+  }
+
+  const innerContent = artifactContent.substring(searchStart, searchEnd);
+  // Remove trailing newline from original for comparison
+  const originalTrimmed = original.replace(/\n$/, '');
+  const relativeIndex = innerContent.indexOf(originalTrimmed);
+
+  if (relativeIndex === -1) {
+    return null;
+  }
+
+  const absoluteIndex = artifact.start + searchStart + relativeIndex;
+  const endText = originalText.substring(absoluteIndex + originalTrimmed.length);
+  const hasTrailingNewline = endText.startsWith('\n');
+
+  const updatedText =
+    originalText.substring(0, absoluteIndex) + updated + (hasTrailingNewline ? '' : '\n') + endText;
+
+  return updatedText.replace(/\n+(?=```\n:::)/g, '\n');
+};
+
+module.exports = {
+  ARTIFACT_START,
+  ARTIFACT_END,
+  findAllArtifacts,
+  replaceArtifactContent,
+};

api/server/services/Artifacts/update.spec.js

@@ -0,0 +1,320 @@
+const {
+  ARTIFACT_START,
+  ARTIFACT_END,
+  findAllArtifacts,
+  replaceArtifactContent,
+} = require('./update');
+
+const createArtifactText = (options = {}) => {
+  const { content = '', wrapCode = true, isClosed = true, prefix = '', suffix = '' } = options;
+
+  const codeBlock = wrapCode ? '```\n' + content + '\n```' : content;
+  const end = isClosed ? `\n${ARTIFACT_END}` : '';
+
+  return `${ARTIFACT_START}${prefix}\n${codeBlock}${end}${suffix}`;
+};
+
+describe('findAllArtifacts', () => {
+  test('should return empty array for message with no artifacts', () => {
+    const message = {
+      content: [
+        {
+          type: 'text',
+          text: 'No artifacts here',
+        },
+      ],
+    };
+    expect(findAllArtifacts(message)).toEqual([]);
+  });
+
+  test('should find artifacts in content parts', () => {
+    const message = {
+      content: [
+        { type: 'text', text: createArtifactText({ content: 'content1' }) },
+        { type: 'text', text: createArtifactText({ content: 'content2' }) },
+      ],
+    };
+
+    const result = findAllArtifacts(message);
+    expect(result).toHaveLength(2);
+    expect(result[0].source).toBe('content');
+    expect(result[1].partIndex).toBe(1);
+  });
+
+  test('should find artifacts in message.text when content is empty', () => {
+    const artifact1 = createArtifactText({ content: 'text1' });
+    const artifact2 = createArtifactText({ content: 'text2' });
+    const message = { text: [artifact1, artifact2].join('\n') };
+
+    const result = findAllArtifacts(message);
+    expect(result).toHaveLength(2);
+    expect(result[0].source).toBe('text');
+  });
+
+  test('should handle unclosed artifacts', () => {
+    const message = {
+      text: createArtifactText({ content: 'unclosed', isClosed: false }),
+    };
+    const result = findAllArtifacts(message);
+    expect(result[0].end).toBe(message.text.length);
+  });
+
+  test('should handle multiple artifacts in single part', () => {
+    const artifact1 = createArtifactText({ content: 'first' });
+    const artifact2 = createArtifactText({ content: 'second' });
+    const message = {
+      content: [
+        {
+          type: 'text',
+          text: [artifact1, artifact2].join('\n'),
+        },
+      ],
+    };
+
+    const result = findAllArtifacts(message);
+    expect(result).toHaveLength(2);
+    expect(result[1].start).toBeGreaterThan(result[0].end);
+  });
+});
+
+describe('replaceArtifactContent', () => {
+  const createTestArtifact = (content, options) => {
+    const text = createArtifactText({ content, ...options });
+    return {
+      start: 0,
+      end: text.length,
+      text,
+      source: 'text',
+    };
+  };
+
+  test('should replace content within artifact boundaries', () => {
+    const original = 'console.log(\'hello\')';
+    const artifact = createTestArtifact(original);
+    const updated = 'console.log(\'updated\')';
+
+    const result = replaceArtifactContent(artifact.text, artifact, original, updated);
+    expect(result).toContain(updated);
+    expect(result).toMatch(ARTIFACT_START);
+    expect(result).toMatch(ARTIFACT_END);
+  });
+
+  test('should return null when original not found', () => {
+    const artifact = createTestArtifact('function test() {}');
+    const result = replaceArtifactContent(artifact.text, artifact, 'missing', 'updated');
+    expect(result).toBeNull();
+  });
+
+  test('should handle dedented content', () => {
+    const original = 'function test() {';
+    const artifact = createTestArtifact(original);
+    const updated = 'function updated() {';
+
+    const result = replaceArtifactContent(artifact.text, artifact, original, updated);
+    expect(result).toContain(updated);
+  });
+
+  test('should preserve text outside artifact', () => {
+    const artifactContent = createArtifactText({ content: 'original' });
+    const fullText = `prefix\n${artifactContent}\nsuffix`;
+    const artifact = createTestArtifact('original', {
+      prefix: 'prefix\n',
+      suffix: '\nsuffix',
+    });
+
+    const result = replaceArtifactContent(fullText, artifact, 'original', 'updated');
+    expect(result).toMatch(/^prefix/);
+    expect(result).toMatch(/suffix$/);
+  });
+
+  test('should handle replacement at artifact boundaries', () => {
+    const original = 'console.log("hello")';
+    const updated = 'console.log("updated")';
+
+    const artifactText = `${ARTIFACT_START}\n${original}\n${ARTIFACT_END}`;
+    const artifact = {
+      start: 0,
+      end: artifactText.length,
+      text: artifactText,
+      source: 'text',
+    };
+
+    const result = replaceArtifactContent(artifactText, artifact, original, updated);
+
+    expect(result).toBe(`${ARTIFACT_START}\n${updated}\n${ARTIFACT_END}`);
+  });
+});
+
+describe('replaceArtifactContent with shared text', () => {
+  test('should replace correct artifact when text is shared', () => {
+    const artifactContent = '    hi    '; // Preserve exact spacing
+    const sharedText = `LOREM IPSUM
+
+:::artifact{identifier="calculator" type="application/vnd.react" title="Calculator"}
+\`\`\`
+${artifactContent}
+\`\`\`
+:::
+
+LOREM IPSUM
+
+:::artifact{identifier="calculator2" type="application/vnd.react" title="Calculator"}
+\`\`\`
+${artifactContent}
+\`\`\`
+:::`;
+
+    const message = { text: sharedText };
+    const artifacts = findAllArtifacts(message);
+    expect(artifacts).toHaveLength(2);
+
+    const targetArtifact = artifacts[1];
+    const updatedContent = '    updated content    ';
+    const result = replaceArtifactContent(
+      sharedText,
+      targetArtifact,
+      artifactContent,
+      updatedContent,
+    );
+
+    // Verify exact matches with preserved formatting
+    expect(result).toContain(artifactContent); // First artifact unchanged
+    expect(result).toContain(updatedContent); // Second artifact updated
+    expect(result.indexOf(updatedContent)).toBeGreaterThan(result.indexOf(artifactContent));
+  });
+
+  const codeExample = `
+function greetPerson(name) {
+  return \`Hello, \${name}! Welcome to JavaScript programming.\`;
+}
+
+const personName = "Alice";
+const greeting = greetPerson(personName);
+console.log(greeting);`;
+
+  test('should handle random number of artifacts in content array', () => {
+    const numArtifacts = 5; // Fixed number for predictability
+    const targetIndex = 2; // Fixed target for predictability
+
+    // Create content array with multiple parts
+    const contentParts = Array.from({ length: numArtifacts }, (_, i) => ({
+      type: 'text',
+      text: createArtifactText({
+        content: `content-${i}`,
+        wrapCode: true,
+        prefix: i > 0 ? '\n' : '',
+      }),
+    }));
+
+    const message = { content: contentParts };
+    const artifacts = findAllArtifacts(message);
+    expect(artifacts).toHaveLength(numArtifacts);
+
+    const targetArtifact = artifacts[targetIndex];
+    const originalContent = `content-${targetIndex}`;
+    const updatedContent = 'updated-content';
+
+    const result = replaceArtifactContent(
+      contentParts[targetIndex].text,
+      targetArtifact,
+      originalContent,
+      updatedContent,
+    );
+
+    // Verify the specific content was updated
+    expect(result).toContain(updatedContent);
+    expect(result).not.toContain(originalContent);
+    expect(result).toMatch(
+      new RegExp(`${ARTIFACT_START}.*${updatedContent}.*${ARTIFACT_END}`, 's'),
+    );
+  });
+
+  test('should handle artifacts with identical content but different metadata in content array', () => {
+    const contentParts = [
+      {
+        type: 'text',
+        text: createArtifactText({
+          wrapCode: true,
+          content: codeExample,
+          prefix: '{id="1", title="First"}',
+        }),
+      },
+      {
+        type: 'text',
+        text: createArtifactText({
+          wrapCode: true,
+          content: codeExample,
+          prefix: '{id="2", title="Second"}',
+        }),
+      },
+    ];
+
+    const message = { content: contentParts };
+    const artifacts = findAllArtifacts(message);
+
+    // Target second artifact
+    const targetArtifact = artifacts[1];
+    const result = replaceArtifactContent(
+      contentParts[1].text,
+      targetArtifact,
+      codeExample,
+      'updated content',
+    );
+    expect(result).toMatch(/id="2".*updated content/s);
+    expect(result).toMatch(new RegExp(`${ARTIFACT_START}.*updated content.*${ARTIFACT_END}`, 's'));
+  });
+
+  test('should handle empty content in artifact without code blocks', () => {
+    const artifactText = `${ARTIFACT_START}\n\n${ARTIFACT_END}`;
+    const artifact = {
+      start: 0,
+      end: artifactText.length,
+      text: artifactText,
+      source: 'text',
+    };
+
+    const result = replaceArtifactContent(artifactText, artifact, '', 'new content');
+    expect(result).toBe(`${ARTIFACT_START}\nnew content\n${ARTIFACT_END}`);
+  });
+
+  test('should handle empty content in artifact with code blocks', () => {
+    const artifactText = createArtifactText({ content: '' });
+    const artifact = {
+      start: 0,
+      end: artifactText.length,
+      text: artifactText,
+      source: 'text',
+    };
+
+    const result = replaceArtifactContent(artifactText, artifact, '', 'new content');
+    expect(result).toMatch(/```\nnew content\n```/);
+  });
+
+  test('should handle content with trailing newline in code blocks', () => {
+    const contentWithNewline = 'console.log("test")\n';
+    const message = {
+      text: `Some prefix text\n${createArtifactText({
+        content: contentWithNewline,
+      })}\nSome suffix text`,
+    };
+
+    const artifacts = findAllArtifacts(message);
+    expect(artifacts).toHaveLength(1);
+
+    const result = replaceArtifactContent(
+      message.text,
+      artifacts[0],
+      contentWithNewline,
+      'updated content',
+    );
+
+    // Should update the content and preserve artifact structure
+    expect(result).toContain('```\nupdated content\n```');
+    // Should preserve surrounding text
+    expect(result).toMatch(/^Some prefix text\n/);
+    expect(result).toMatch(/\nSome suffix text$/);
+    // Should not have extra newlines
+    expect(result).not.toContain('\n\n```');
+    expect(result).not.toContain('```\n\n');
+  });
+});

api/server/services/AuthService.js

@@ -35,13 +35,14 @@ const genericVerificationMessage = 'Please check your email to verify your email
 /**
  * Logout user
  *
- * @param {String} userId
- * @param {*} refreshToken
+ * @param {ServerRequest} req
+ * @param {string} refreshToken
  * @returns
  */
-const logoutUser = async (userId, refreshToken) => {
+const logoutUser = async (req, refreshToken) => {
   try {
-    const session = await findSession({ userId: userId, refreshToken: refreshToken });
+    const userId = req.user._id;
+    const session = await findSession({ userId: userId, refreshToken });
 
     if (session) {
       try {
@@ -52,6 +53,12 @@ const logoutUser = async (userId, refreshToken) => {
       }
     }
 
+    try {
+      req.session.destroy();
+    } catch (destroyErr) {
+      logger.error('[logoutUser] Failed to destroy session.', destroyErr);
+    }
+
     return { status: 200, message: 'Logout successful' };
   } catch (err) {
     return { status: 500, message: err.message };
@@ -108,31 +115,46 @@ const sendVerificationEmail = async (user) => {
  */
 const verifyEmail = async (req) => {
   const { email, token } = req.body;
-  let emailVerificationData = await findToken({ email: decodeURIComponent(email) });
+  const decodedEmail = decodeURIComponent(email);
+
+  const user = await findUser({ email: decodedEmail }, 'email _id emailVerified');
+
+  if (!user) {
+    logger.warn(`[verifyEmail] [User not found] [Email: ${decodedEmail}]`);
+    return new Error('User not found');
+  }
+
+  if (user.emailVerified) {
+    logger.info(`[verifyEmail] Email already verified [Email: ${decodedEmail}]`);
+    return { message: 'Email already verified', status: 'success' };
+  }
+
+  let emailVerificationData = await findToken({ email: decodedEmail });
 
   if (!emailVerificationData) {
-    logger.warn(`[verifyEmail] [No email verification data found] [Email: ${email}]`);
+    logger.warn(`[verifyEmail] [No email verification data found] [Email: ${decodedEmail}]`);
     return new Error('Invalid or expired password reset token');
   }
 
   const isValid = bcrypt.compareSync(token, emailVerificationData.token);
 
   if (!isValid) {
-    logger.warn(`[verifyEmail] [Invalid or expired email verification token] [Email: ${email}]`);
+    logger.warn(
+      `[verifyEmail] [Invalid or expired email verification token] [Email: ${decodedEmail}]`,
+    );
     return new Error('Invalid or expired email verification token');
   }
 
   const updatedUser = await updateUser(emailVerificationData.userId, { emailVerified: true });
   if (!updatedUser) {
-    logger.warn(`[verifyEmail] [User not found] [Email: ${email}]`);
-    return new Error('User not found');
+    logger.warn(`[verifyEmail] [User update failed] [Email: ${decodedEmail}]`);
+    return new Error('Failed to update user verification status');
   }
 
   await deleteTokens({ token: emailVerificationData.token });
-  logger.info(`[verifyEmail] Email verification successful. [Email: ${email}]`);
-  return { message: 'Email verification was successful' };
+  logger.info(`[verifyEmail] Email verification successful [Email: ${decodedEmail}]`);
+  return { message: 'Email verification was successful', status: 'success' };
 };
-
 /**
  * Register a new user.
  * @param {MongoUser} user <email, password, name, username>

api/server/services/Config/EndpointService.js

@@ -8,7 +8,6 @@ const {
   AZURE_API_KEY: azureOpenAIApiKey,
   ANTHROPIC_API_KEY: anthropicApiKey,
   CHATGPT_TOKEN: chatGPTToken,
-  BINGAI_TOKEN: bingToken,
   PLUGINS_USE_AZURE,
   GOOGLE_KEY: googleKey,
   OPENAI_REVERSE_PROXY,
@@ -30,7 +29,6 @@ module.exports = {
     useAzurePlugins,
     userProvidedOpenAI,
     googleKey,
-    [EModelEndpoint.bingAI]: generateConfig(bingToken),
     [EModelEndpoint.anthropic]: generateConfig(anthropicApiKey),
     [EModelEndpoint.chatGPTBrowser]: generateConfig(chatGPTToken),
     [EModelEndpoint.openAI]: generateConfig(openAIApiKey, OPENAI_REVERSE_PROXY),

api/server/services/Config/loadAsyncEndpoints.js

@@ -1,6 +1,4 @@
 const { EModelEndpoint } = require('librechat-data-provider');
-const { addOpenAPISpecs } = require('~/app/clients/tools/util/addOpenAPISpecs');
-const { availableTools } = require('~/app/clients/tools');
 const { isUserProvided } = require('~/server/utils');
 const { config } = require('./EndpointService');
 
@@ -28,22 +26,12 @@ async function loadAsyncEndpoints(req) {
     }
   }
 
-  const tools = await addOpenAPISpecs(availableTools);
-  function transformToolsToMap(tools) {
-    return tools.reduce((map, obj) => {
-      map[obj.pluginKey] = obj.name;
-      return map;
-    }, {});
-  }
-  const plugins = transformToolsToMap(tools);
-
   const google = serviceKey || googleKey ? { userProvide: googleUserProvides } : false;
 
   const useAzure = req.app.locals[EModelEndpoint.azureOpenAI]?.plugins;
   const gptPlugins =
     useAzure || openAIApiKey || azureOpenAIApiKey
       ? {
-        plugins,
         availableAgents: ['classic', 'functions'],
         userProvide: useAzure ? false : userProvidedOpenAI,
         userProvideURL: useAzure

api/server/services/Config/loadDefaultEConfig.js

@@ -9,7 +9,7 @@ const { config } = require('./EndpointService');
  */
 async function loadDefaultEndpointsConfig(req) {
   const { google, gptPlugins } = await loadAsyncEndpoints(req);
-  const { assistants, azureAssistants, bingAI, azureOpenAI, chatGPTBrowser } = config;
+  const { assistants, azureAssistants, azureOpenAI, chatGPTBrowser } = config;
 
   const enabledEndpoints = getEnabledEndpoints();
 
@@ -20,7 +20,6 @@ async function loadDefaultEndpointsConfig(req) {
     [EModelEndpoint.azureAssistants]: azureAssistants,
     [EModelEndpoint.azureOpenAI]: azureOpenAI,
     [EModelEndpoint.google]: google,
-    [EModelEndpoint.bingAI]: bingAI,
     [EModelEndpoint.chatGPTBrowser]: chatGPTBrowser,
     [EModelEndpoint.gptPlugins]: gptPlugins,
     [EModelEndpoint.anthropic]: config[EModelEndpoint.anthropic],

api/server/services/Config/loadDefaultModels.js

@@ -35,7 +35,6 @@ async function loadDefaultModels(req) {
     [EModelEndpoint.anthropic]: anthropic,
     [EModelEndpoint.gptPlugins]: gptPlugins,
     [EModelEndpoint.azureOpenAI]: azureOpenAI,
-    [EModelEndpoint.bingAI]: ['BingAI', 'Sydney'],
     [EModelEndpoint.chatGPTBrowser]: chatGPTBrowser,
     [EModelEndpoint.assistants]: assistants,
     [EModelEndpoint.azureAssistants]: azureAssistants,

api/server/services/Endpoints/agents/initialize.js

@@ -13,6 +13,7 @@ const getBedrockOptions = require('~/server/services/Endpoints/bedrock/options')
 const initOpenAI = require('~/server/services/Endpoints/openAI/initialize');
 const initCustom = require('~/server/services/Endpoints/custom/initialize');
 const initGoogle = require('~/server/services/Endpoints/google/initialize');
+const generateArtifactsPrompt = require('~/app/clients/prompts/artifacts');
 const { getCustomEndpointConfig } = require('~/server/services/Config');
 const { loadAgentTools } = require('~/server/services/ToolService');
 const AgentClient = require('~/server/controllers/agents/client');
@@ -72,6 +73,16 @@ const primeResources = async (_attachments, _tool_resources) => {
   }
 };
 
+/**
+ * @param {object} params
+ * @param {ServerRequest} params.req
+ * @param {ServerResponse} params.res
+ * @param {Agent} params.agent
+ * @param {object} [params.endpointOption]
+ * @param {AgentToolResources} [params.tool_resources]
+ * @param {boolean} [params.isInitialAgent]
+ * @returns {Promise<Agent>}
+ */
 const initializeAgentOptions = async ({
   req,
   res,
@@ -82,6 +93,7 @@ const initializeAgentOptions = async ({
 }) => {
   const { tools, toolContextMap } = await loadAgentTools({
     req,
+    res,
     agent,
     tool_resources,
   });
@@ -131,6 +143,13 @@ const initializeAgentOptions = async ({
     agent.model_parameters.model = agent.model;
   }
 
+  if (typeof agent.artifacts === 'string' && agent.artifacts !== '') {
+    agent.additional_instructions = generateArtifactsPrompt({
+      endpoint: agent.provider,
+      artifacts: agent.artifacts,
+    });
+  }
+
   const tokensModel =
     agent.provider === EModelEndpoint.azureOpenAI ? agent.model : agent.model_parameters.model;
 

api/server/services/Endpoints/assistants/build.js

@@ -3,7 +3,7 @@ const generateArtifactsPrompt = require('~/app/clients/prompts/artifacts');
 const { getAssistant } = require('~/models/Assistant');
 
 const buildOptions = async (endpoint, parsedBody) => {
-  // eslint-disable-next-line no-unused-vars
+
   const { promptPrefix, assistant_id, iconURL, greeting, spec, artifacts, ...modelOptions } =
     parsedBody;
   const endpointOption = removeNullishValues({

api/server/services/Endpoints/azureAssistants/build.js

@@ -3,7 +3,7 @@ const generateArtifactsPrompt = require('~/app/clients/prompts/artifacts');
 const { getAssistant } = require('~/models/Assistant');
 
 const buildOptions = async (endpoint, parsedBody) => {
-  // eslint-disable-next-line no-unused-vars
+
   const { promptPrefix, assistant_id, iconURL, greeting, spec, artifacts, ...modelOptions } =
     parsedBody;
   const endpointOption = removeNullishValues({

api/server/services/Endpoints/google/build.js

@@ -11,6 +11,7 @@ const buildOptions = (endpoint, parsedBody) => {
     greeting,
     spec,
     artifacts,
+    maxContextTokens,
     ...modelOptions
   } = parsedBody;
   const endpointOption = removeNullishValues({
@@ -22,6 +23,7 @@ const buildOptions = (endpoint, parsedBody) => {
     iconURL,
     greeting,
     spec,
+    maxContextTokens,
     modelOptions,
   });
 

api/server/services/Endpoints/google/llm.js

@@ -1,21 +1,42 @@
 const { Providers } = require('@librechat/agents');
 const { AuthKeys } = require('librechat-data-provider');
+const { isEnabled } = require('~/server/utils');
 
-// Example internal constant from your code
-const EXCLUDED_GENAI_MODELS = /gemini-(?:1\.0|1-0|pro)/;
+function getThresholdMapping(model) {
+  const gemini1Pattern = /gemini-(1\.0|1\.5|pro$|1\.0-pro|1\.5-pro|1\.5-flash-001)/;
+  const restrictedPattern = /(gemini-(1\.5-flash-8b|2\.0|exp)|learnlm)/;
+
+  if (gemini1Pattern.test(model)) {
+    return (value) => {
+      if (value === 'OFF') {
+        return 'BLOCK_NONE';
+      }
+      return value;
+    };
+  }
+
+  if (restrictedPattern.test(model)) {
+    return (value) => {
+      if (value === 'OFF' || value === 'HARM_BLOCK_THRESHOLD_UNSPECIFIED') {
+        return 'BLOCK_NONE';
+      }
+      return value;
+    };
+  }
+
+  return (value) => value;
+}
 
 /**
  *
- * @param {boolean} isGemini2
- * @returns {Array<{category: string, threshold: string}>}
+ * @param {string} model
+ * @returns {Array<{category: string, threshold: string}> | undefined}
  */
-function getSafetySettings(isGemini2) {
-  const mapThreshold = (value) => {
-    if (isGemini2 && value === 'BLOCK_NONE') {
-      return 'OFF';
-    }
-    return value;
-  };
+function getSafetySettings(model) {
+  if (isEnabled(process.env.GOOGLE_EXCLUDE_SAFETY_SETTINGS)) {
+    return undefined;
+  }
+  const mapThreshold = getThresholdMapping(model);
 
   return [
     {
@@ -87,23 +108,13 @@ function getLLMConfig(credentials, options = {}) {
     maxRetries: 2,
   };
 
-  const isGemini2 = llmConfig.model.includes('gemini-2.0');
-  const isGenerativeModel = llmConfig.model.includes('gemini');
-  const isChatModel = !isGenerativeModel && llmConfig.model.includes('chat');
-  const isTextModel = !isGenerativeModel && !isChatModel && /code|text/.test(llmConfig.model);
-
-  llmConfig.safetySettings = getSafetySettings(isGemini2);
+  /** Used only for Safety Settings */
+  llmConfig.safetySettings = getSafetySettings(llmConfig.model);
 
   let provider;
 
-  if (project_id && isTextModel) {
+  if (project_id) {
     provider = Providers.VERTEXAI;
-  } else if (project_id && isChatModel) {
-    provider = Providers.VERTEXAI;
-  } else if (project_id) {
-    provider = Providers.VERTEXAI;
-  } else if (!EXCLUDED_GENAI_MODELS.test(llmConfig.model)) {
-    provider = Providers.GOOGLE;
   } else {
     provider = Providers.GOOGLE;
   }
@@ -165,4 +176,5 @@ function getLLMConfig(credentials, options = {}) {
 
 module.exports = {
   getLLMConfig,
+  getSafetySettings,
 };