Merge pull request #4302 from janhq/dev

Release cut 0.5.12

.github/ISSUE_TEMPLATE/roadmap.md

@@ -1,3 +1,12 @@
+---
+name: Roadmap
+about: Plan Roadmap items with subtasks
+title: 'roadmap: '
+labels: 'type: planning'
+assignees: ''
+
+---
+
 ## Goal
 
 ## Tasklist

.github/workflows/jan-electron-build-beta.yml

@@ -70,6 +70,8 @@ jobs:
     permissions:
       contents: write
     steps:
+      - name: Getting the repo
+        uses: actions/checkout@v3
       - name: Sync temp to latest
         run: |
           # sync temp-beta to beta by copy files that are different or new

.github/workflows/jan-electron-build-nightly.yml

@@ -12,6 +12,8 @@ on:
           - none
           - aws-s3
         default: none
+  pull_request_review:
+    types: [submitted]
 
 jobs:
   set-public-provider:
@@ -33,6 +35,9 @@ jobs:
             elif [ "${{ github.event_name }}" == "push" ]; then
               echo "::set-output name=public_provider::aws-s3"
               echo "::set-output name=ref::${{ github.ref }}"
+            elif [ "${{ github.event_name }}" == "pull_request_review" ]; then
+              echo "::set-output name=public_provider::none"
+              echo "::set-output name=ref::${{ github.ref }}"
             else
               echo "::set-output name=public_provider::none"
               echo "::set-output name=ref::${{ github.ref }}"
@@ -116,3 +121,24 @@ jobs:
       build_reason: Manual
       push_to_branch: dev
       new_version: ${{ needs.get-update-version.outputs.new_version }}
+
+
+  comment-pr-build-url:
+    needs: [build-macos, build-windows-x64, build-linux-x64, get-update-version, set-public-provider, sync-temp-to-latest]
+    runs-on: ubuntu-latest
+    if: github.event_name == 'pull_request_review'
+    steps:
+      - name: Set up GitHub CLI
+        run: |
+          curl -sSL https://github.com/cli/cli/releases/download/v2.33.0/gh_2.33.0_linux_amd64.tar.gz | tar xz
+          sudo cp gh_2.33.0_linux_amd64/bin/gh /usr/local/bin/
+
+      - name: Comment build URL on PR
+        env:
+          GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
+        run: |
+          PR_URL=${{ github.event.pull_request.html_url }}
+          RUN_ID=${{ github.run_id }}
+          COMMENT="This is the build for this pull request. You can download it from the Artifacts section here: [Build URL](https://github.com/${{ github.repository }}/actions/runs/${RUN_ID})."
+          gh pr comment $PR_URL --body "$COMMENT"
+    

.github/workflows/publish-npm-core.yml

@@ -0,0 +1,53 @@
+name: Publish plugin models Package to npmjs
+on:
+  push:
+    tags: ["v[0-9]+.[0-9]+.[0-9]+-core"]
+    paths: ["core/**"]
+  pull_request:
+    paths: ["core/**"]
+jobs:
+  build-and-publish-plugins:
+    environment: production
+    runs-on: ubuntu-latest
+    steps:
+      - uses: actions/checkout@v4
+        with:
+          fetch-depth: "0"
+          token: ${{ secrets.PAT_SERVICE_ACCOUNT }}
+
+      - name: Install jq
+        uses: dcarbone/install-jq-action@v2.0.1
+
+      - name: Extract tag name without v prefix
+        id: get_version
+        run: echo "VERSION=${GITHUB_REF#refs/tags/v}" >> $GITHUB_ENV && echo "::set-output name=version::${GITHUB_REF#refs/tags/v}"
+        env:
+          GITHUB_REF: ${{ github.ref }}
+
+      - name: "Get Semantic Version from tag"
+        if: github.event_name == 'push'
+        run: |
+          # Get the tag from the event
+          tag=${GITHUB_REF#refs/tags/v}
+          # remove the -core suffix
+          new_version=$(echo $tag | sed -n 's/-core//p')
+          echo $new_version
+          # Replace the old version with the new version in package.json
+          jq --arg version "$new_version" '.version = $version' core/package.json > /tmp/package.json && mv /tmp/package.json core/package.json
+
+          # Print the new version
+          echo "Updated package.json version to: $new_version"
+          cat core/package.json
+
+      # Setup .npmrc file to publish to npm
+      - uses: actions/setup-node@v3
+        with:
+          node-version: "20.x"
+          registry-url: "https://registry.npmjs.org"
+
+      - run: cd core && yarn install && yarn build
+
+      - run: cd core && yarn publish --access public
+        if: github.event_name == 'push'
+        env:
+          NODE_AUTH_TOKEN: ${{ secrets.NPM_TOKEN }}

.github/workflows/publish-npm-joi.yml

@@ -0,0 +1,53 @@
+name: Publish plugin models Package to npmjs
+on:
+  push:
+    tags: ["v[0-9]+.[0-9]+.[0-9]+-joi"]
+    paths: ["joi/**"]
+  pull_request:
+    paths: ["joi/**"]
+jobs:
+  build-and-publish-plugins:
+    environment: production
+    runs-on: ubuntu-latest
+    steps:
+      - uses: actions/checkout@v4
+        with:
+          fetch-depth: "0"
+          token: ${{ secrets.PAT_SERVICE_ACCOUNT }}
+
+      - name: Install jq
+        uses: dcarbone/install-jq-action@v2.0.1
+
+      - name: Extract tag name without v prefix
+        id: get_version
+        run: echo "VERSION=${GITHUB_REF#refs/tags/v}" >> $GITHUB_ENV && echo "::set-output name=version::${GITHUB_REF#refs/tags/v}"
+        env:
+          GITHUB_REF: ${{ github.ref }}
+
+      - name: "Get Semantic Version from tag"
+        if: github.event_name == 'push'
+        run: |
+          # Get the tag from the event
+          tag=${GITHUB_REF#refs/tags/v}
+          # remove the -joi suffix
+          new_version=$(echo $tag | sed -n 's/-joi//p')
+          echo $new_version
+          # Replace the old version with the new version in package.json
+          jq --arg version "$new_version" '.version = $version' joi/package.json > /tmp/package.json && mv /tmp/package.json joi/package.json
+
+          # Print the new version
+          echo "Updated package.json version to: $new_version"
+          cat joi/package.json
+
+      # Setup .npmrc file to publish to npm
+      - uses: actions/setup-node@v3
+        with:
+          node-version: "20.x"
+          registry-url: "https://registry.npmjs.org"
+
+      - run: cd joi && yarn install && yarn build
+
+      - run: cd joi && yarn publish --access public
+        if: github.event_name == 'push'
+        env:
+          NODE_AUTH_TOKEN: ${{ secrets.NPM_TOKEN }}

.github/workflows/template-build-linux-x64.yml

@@ -111,8 +111,10 @@ jobs:
           GH_TOKEN: ${{ secrets.GITHUB_TOKEN }}
           AWS_ACCESS_KEY_ID: ${{ secrets.DELTA_AWS_ACCESS_KEY_ID }}
           AWS_SECRET_ACCESS_KEY: ${{ secrets.DELTA_AWS_SECRET_ACCESS_KEY }}
-          AWS_EC2_METADATA_DISABLED: "true"
-          AWS_MAX_ATTEMPTS: "5"
+          AWS_EC2_METADATA_DISABLED: 'true'
+          AWS_MAX_ATTEMPTS: '5'
+          POSTHOG_KEY: ${{ secrets.POSTHOG_KEY }}
+          POSTHOG_HOST: ${{ secrets.POSTHOG_HOST }}
 
       - name: Build and publish app to github
         if: github.event_name == 'push' && startsWith(github.ref, 'refs/tags/') && inputs.public_provider == 'github' && inputs.beta == false
@@ -122,6 +124,8 @@ jobs:
           GH_TOKEN: ${{ secrets.GITHUB_TOKEN }}
           ANALYTICS_ID: ${{ secrets.JAN_APP_UMAMI_PROJECT_API_KEY }}
           ANALYTICS_HOST: ${{ secrets.JAN_APP_UMAMI_URL }}
+          POSTHOG_KEY: ${{ secrets.POSTHOG_KEY }}
+          POSTHOG_HOST: ${{ secrets.POSTHOG_HOST }}
 
       - name: Build and publish app to github
         if: github.event_name == 'push' && startsWith(github.ref, 'refs/tags/') && inputs.public_provider == 'github' && inputs.beta == true
@@ -131,8 +135,10 @@ jobs:
           GH_TOKEN: ${{ secrets.GITHUB_TOKEN }}
           AWS_ACCESS_KEY_ID: ${{ secrets.DELTA_AWS_ACCESS_KEY_ID }}
           AWS_SECRET_ACCESS_KEY: ${{ secrets.DELTA_AWS_SECRET_ACCESS_KEY }}
-          AWS_EC2_METADATA_DISABLED: "true"
-          AWS_MAX_ATTEMPTS: "5"
+          AWS_EC2_METADATA_DISABLED: 'true'
+          AWS_MAX_ATTEMPTS: '5'
+          POSTHOG_KEY: ${{ secrets.POSTHOG_KEY }}
+          POSTHOG_HOST: ${{ secrets.POSTHOG_HOST }}
 
       - name: Upload Artifact .deb file
         if: inputs.public_provider != 'github'

.github/workflows/template-build-macos.yml

@@ -140,18 +140,20 @@ jobs:
           fi
         env:
           GH_TOKEN: ${{ secrets.GITHUB_TOKEN }}
-          CSC_LINK: "/tmp/codesign.p12"
+          CSC_LINK: '/tmp/codesign.p12'
           CSC_KEY_PASSWORD: ${{ secrets.CODE_SIGN_P12_PASSWORD }}
-          CSC_IDENTITY_AUTO_DISCOVERY: "true"
+          CSC_IDENTITY_AUTO_DISCOVERY: 'true'
           APPLE_ID: ${{ secrets.APPLE_ID }}
           APPLE_APP_SPECIFIC_PASSWORD: ${{ secrets.APPLE_APP_SPECIFIC_PASSWORD }}
-          APP_PATH: "."
+          APP_PATH: '.'
           DEVELOPER_ID: ${{ secrets.DEVELOPER_ID }}
           AWS_ACCESS_KEY_ID: ${{ secrets.DELTA_AWS_ACCESS_KEY_ID }}
           AWS_SECRET_ACCESS_KEY: ${{ secrets.DELTA_AWS_SECRET_ACCESS_KEY }}
           AWS_DEFAULT_REGION: auto
-          AWS_EC2_METADATA_DISABLED: "true"
-          AWS_MAX_ATTEMPTS: "5"
+          AWS_EC2_METADATA_DISABLED: 'true'
+          AWS_MAX_ATTEMPTS: '5'
+          POSTHOG_KEY: ${{ secrets.POSTHOG_KEY }}
+          POSTHOG_HOST: ${{ secrets.POSTHOG_HOST }}
 
       - name: Build and publish app to github
         if: github.event_name == 'push' && startsWith(github.ref, 'refs/tags/') && inputs.public_provider == 'github' && inputs.beta == false
@@ -159,15 +161,17 @@ jobs:
           make build-and-publish
         env:
           GH_TOKEN: ${{ secrets.GITHUB_TOKEN }}
-          CSC_LINK: "/tmp/codesign.p12"
+          CSC_LINK: '/tmp/codesign.p12'
           CSC_KEY_PASSWORD: ${{ secrets.CODE_SIGN_P12_PASSWORD }}
-          CSC_IDENTITY_AUTO_DISCOVERY: "true"
+          CSC_IDENTITY_AUTO_DISCOVERY: 'true'
           APPLE_ID: ${{ secrets.APPLE_ID }}
           APPLE_APP_SPECIFIC_PASSWORD: ${{ secrets.APPLE_APP_SPECIFIC_PASSWORD }}
-          APP_PATH: "."
+          APP_PATH: '.'
           DEVELOPER_ID: ${{ secrets.DEVELOPER_ID }}
           ANALYTICS_ID: ${{ secrets.JAN_APP_UMAMI_PROJECT_API_KEY }}
           ANALYTICS_HOST: ${{ secrets.JAN_APP_UMAMI_URL }}
+          POSTHOG_KEY: ${{ secrets.POSTHOG_KEY }}
+          POSTHOG_HOST: ${{ secrets.POSTHOG_HOST }}
 
       - name: Build and publish app to github
         if: github.event_name == 'push' && startsWith(github.ref, 'refs/tags/') && inputs.public_provider == 'github' && inputs.beta == true
@@ -175,18 +179,20 @@ jobs:
           make build-and-publish
         env:
           GH_TOKEN: ${{ secrets.GITHUB_TOKEN }}
-          CSC_LINK: "/tmp/codesign.p12"
+          CSC_LINK: '/tmp/codesign.p12'
           CSC_KEY_PASSWORD: ${{ secrets.CODE_SIGN_P12_PASSWORD }}
-          CSC_IDENTITY_AUTO_DISCOVERY: "true"
+          CSC_IDENTITY_AUTO_DISCOVERY: 'true'
           APPLE_ID: ${{ secrets.APPLE_ID }}
           APPLE_APP_SPECIFIC_PASSWORD: ${{ secrets.APPLE_APP_SPECIFIC_PASSWORD }}
-          APP_PATH: "."
+          APP_PATH: '.'
           DEVELOPER_ID: ${{ secrets.DEVELOPER_ID }}
           AWS_ACCESS_KEY_ID: ${{ secrets.DELTA_AWS_ACCESS_KEY_ID }}
           AWS_SECRET_ACCESS_KEY: ${{ secrets.DELTA_AWS_SECRET_ACCESS_KEY }}
           AWS_DEFAULT_REGION: auto
-          AWS_EC2_METADATA_DISABLED: "true"
-          AWS_MAX_ATTEMPTS: "5"
+          AWS_EC2_METADATA_DISABLED: 'true'
+          AWS_MAX_ATTEMPTS: '5'
+          POSTHOG_KEY: ${{ secrets.POSTHOG_KEY }}
+          POSTHOG_HOST: ${{ secrets.POSTHOG_HOST }}
 
       - name: Upload Artifact
         if: inputs.public_provider != 'github'

.github/workflows/template-build-windows-x64.yml

@@ -149,8 +149,10 @@ jobs:
           AWS_ACCESS_KEY_ID: ${{ secrets.DELTA_AWS_ACCESS_KEY_ID }}
           AWS_SECRET_ACCESS_KEY: ${{ secrets.DELTA_AWS_SECRET_ACCESS_KEY }}
           AWS_DEFAULT_REGION: auto
-          AWS_EC2_METADATA_DISABLED: "true"
-          AWS_MAX_ATTEMPTS: "5"
+          AWS_EC2_METADATA_DISABLED: 'true'
+          AWS_MAX_ATTEMPTS: '5'
+          POSTHOG_KEY: ${{ secrets.POSTHOG_KEY }}
+          POSTHOG_HOST: ${{ secrets.POSTHOG_HOST }}
 
       - name: Build app and publish app to github
         if: github.event_name == 'push' && startsWith(github.ref, 'refs/tags/') && inputs.public_provider == 'github' && inputs.beta == false
@@ -165,6 +167,8 @@ jobs:
           AZURE_TENANT_ID: ${{ secrets.AZURE_TENANT_ID }}
           AZURE_CLIENT_SECRET: ${{ secrets.AZURE_CLIENT_SECRET }}
           AZURE_CERT_NAME: homebrewltd
+          POSTHOG_KEY: ${{ secrets.POSTHOG_KEY }}
+          POSTHOG_HOST: ${{ secrets.POSTHOG_HOST }}
 
       - name: Build app and publish app to github
         if: github.event_name == 'push' && startsWith(github.ref, 'refs/tags/') && inputs.public_provider == 'github' && inputs.beta == true
@@ -175,14 +179,16 @@ jobs:
           AWS_ACCESS_KEY_ID: ${{ secrets.DELTA_AWS_ACCESS_KEY_ID }}
           AWS_SECRET_ACCESS_KEY: ${{ secrets.DELTA_AWS_SECRET_ACCESS_KEY }}
           AWS_DEFAULT_REGION: auto
-          AWS_EC2_METADATA_DISABLED: "true"
-          AWS_MAX_ATTEMPTS: "5"
+          AWS_EC2_METADATA_DISABLED: 'true'
+          AWS_MAX_ATTEMPTS: '5'
           AZURE_KEY_VAULT_URI: ${{ secrets.AZURE_KEY_VAULT_URI }}
           AZURE_CLIENT_ID: ${{ secrets.AZURE_CLIENT_ID }}
           AZURE_TENANT_ID: ${{ secrets.AZURE_TENANT_ID }}
           AZURE_CLIENT_SECRET: ${{ secrets.AZURE_CLIENT_SECRET }}
           # AZURE_CERT_NAME: ${{ secrets.AZURE_CERT_NAME }}
           AZURE_CERT_NAME: homebrewltd
+          POSTHOG_KEY: ${{ secrets.POSTHOG_KEY }}
+          POSTHOG_HOST: ${{ secrets.POSTHOG_HOST }}
 
       - name: Upload Artifact
         if: inputs.public_provider != 'github'
@@ -190,4 +196,3 @@ jobs:
         with:
           name: jan-win-x64-${{ inputs.new_version }}
           path: ./electron/dist/*.exe
-

README.md

@@ -47,7 +47,7 @@ From PCs to multi-GPU clusters, Jan & Cortex supports universal architectures:
   <tr style="text-align:center">
     <td style="text-align:center"><b>Version Type</b></td>
     <td style="text-align:center"><b>Windows</b></td>
-    <td colspan="2" style="text-align:center"><b>MacOS</b></td>
+    <td style="text-align:center"><b>MacOS Universal</b></td>
     <td colspan="2" style="text-align:center"><b>Linux</b></td>
   </tr>
   <tr style="text-align:center">
@@ -59,15 +59,9 @@ From PCs to multi-GPU clusters, Jan & Cortex supports universal architectures:
       </a>
     </td>
     <td style="text-align:center">
-      <a href='https://app.jan.ai/download/latest/mac-x64'>
+      <a href='https://app.jan.ai/download/latest/mac-universal'>
         <img src='https://github.com/janhq/jan/blob/dev/docs/static/img/mac.png' style="height:15px; width: 15px" />
-        <b>Intel</b>
-      </a>
-    </td>
-    <td style="text-align:center">
-      <a href='https://app.jan.ai/download/latest/mac-arm64'>
-        <img src='https://github.com/janhq/jan/blob/dev/docs/static/img/mac.png' style="height:15px; width: 15px" />
-        <b>M1/M2/M3/M4</b>
+        <b>jan.dmg</b>
       </a>
     </td>
     <td style="text-align:center">
@@ -92,15 +86,9 @@ From PCs to multi-GPU clusters, Jan & Cortex supports universal architectures:
       </a>
     </td>
     <td style="text-align:center">
-      <a href='https://app.jan.ai/download/beta/mac-x64'>
+      <a href='https://app.jan.ai/download/beta/mac-universal'>
         <img src='https://github.com/janhq/jan/blob/dev/docs/static/img/mac.png' style="height:15px; width: 15px" />
-        <b>Intel</b>
-      </a>
-    </td>
-    <td style="text-align:center">
-      <a href='https://app.jan.ai/download/beta/mac-arm64'>
-        <img src='https://github.com/janhq/jan/blob/dev/docs/static/img/mac.png' style="height:15px; width: 15px" />
-        <b>M1/M2/M3/M4</b>
+        <b>jan.dmg</b>
       </a>
     </td>
     <td style="text-align:center">
@@ -125,15 +113,9 @@ From PCs to multi-GPU clusters, Jan & Cortex supports universal architectures:
       </a>
     </td>
     <td style="text-align:center">
-      <a href='https://app.jan.ai/download/nightly/mac-x64'>
+      <a href='https://app.jan.ai/download/nightly/mac-universal'>
         <img src='https://github.com/janhq/jan/blob/dev/docs/static/img/mac.png' style="height:15px; width: 15px" />
-        <b>Intel</b>
-      </a>
-    </td>
-    <td style="text-align:center">
-      <a href='https://app.jan.ai/download/nightly/mac-arm64'>
-        <img src='https://github.com/janhq/jan/blob/dev/docs/static/img/mac.png' style="height:15px; width: 15px" />
-        <b>M1/M2/M3/M4</b>
+        <b>jan.dmg</b>
       </a>
     </td>
     <td style="text-align:center">

core/src/browser/extensions/conversational.ts

@@ -1,4 +1,10 @@
-import { Thread, ThreadInterface, ThreadMessage, MessageInterface } from '../../types'
+import {
+  Thread,
+  ThreadInterface,
+  ThreadMessage,
+  MessageInterface,
+  ThreadAssistantInfo,
+} from '../../types'
 import { BaseExtension, ExtensionTypeEnum } from '../extension'
 
 /**
@@ -17,10 +23,21 @@ export abstract class ConversationalExtension
     return ExtensionTypeEnum.Conversational
   }
 
-  abstract getThreads(): Promise<Thread[]>
-  abstract saveThread(thread: Thread): Promise<void>
+  abstract listThreads(): Promise<Thread[]>
+  abstract createThread(thread: Partial<Thread>): Promise<Thread>
+  abstract modifyThread(thread: Thread): Promise<void>
   abstract deleteThread(threadId: string): Promise<void>
-  abstract addNewMessage(message: ThreadMessage): Promise<void>
-  abstract writeMessages(threadId: string, messages: ThreadMessage[]): Promise<void>
-  abstract getAllMessages(threadId: string): Promise<ThreadMessage[]>
+  abstract createMessage(message: Partial<ThreadMessage>): Promise<ThreadMessage>
+  abstract deleteMessage(threadId: string, messageId: string): Promise<void>
+  abstract listMessages(threadId: string): Promise<ThreadMessage[]>
+  abstract getThreadAssistant(threadId: string): Promise<ThreadAssistantInfo>
+  abstract createThreadAssistant(
+    threadId: string,
+    assistant: ThreadAssistantInfo
+  ): Promise<ThreadAssistantInfo>
+  abstract modifyThreadAssistant(
+    threadId: string,
+    assistant: ThreadAssistantInfo
+  ): Promise<ThreadAssistantInfo>
+  abstract modifyMessage(message: ThreadMessage): Promise<ThreadMessage>
 }

core/src/browser/extensions/engines/AIEngine.ts

@@ -2,7 +2,6 @@ import { events } from '../../events'
 import { BaseExtension } from '../../extension'
 import { MessageRequest, Model, ModelEvent } from '../../../types'
 import { EngineManager } from './EngineManager'
-import { ModelManager } from '../../models/manager'
 
 /**
  * Base AIEngine

core/src/browser/extensions/engines/OAIEngine.ts

@@ -71,7 +71,7 @@ export abstract class OAIEngine extends AIEngine {
       return
     }
 
-    const timestamp = Date.now()
+    const timestamp = Date.now() / 1000
     const message: ThreadMessage = {
       id: ulid(),
       thread_id: data.threadId,
@@ -80,8 +80,8 @@ export abstract class OAIEngine extends AIEngine {
       role: ChatCompletionRole.Assistant,
       content: [],
       status: MessageStatus.Pending,
-      created: timestamp,
-      updated: timestamp,
+      created_at: timestamp,
+      completed_at: timestamp,
       object: 'thread.message',
     }
 

core/src/node/api/HttpServer.ts

@@ -1,8 +0,0 @@
-export interface HttpServer {
-  post: (route: string, handler: (req: any, res: any) => Promise<any>) => void
-  get: (route: string, handler: (req: any, res: any) => Promise<any>) => void
-  patch: (route: string, handler: (req: any, res: any) => Promise<any>) => void
-  put: (route: string, handler: (req: any, res: any) => Promise<any>) => void
-  delete: (route: string, handler: (req: any, res: any) => Promise<any>) => void
-  register: (router: any, opts?: any) => void
-}

core/src/node/api/index.test.ts

@@ -1,7 +0,0 @@
-
-import * as restfulV1 from './restful/v1';
-
-it('should re-export from restful/v1', () => {
-  const restfulV1Exports = require('./restful/v1');
-  expect(restfulV1Exports).toBeDefined();
-})

core/src/node/api/index.ts

@@ -1,3 +1 @@
-export * from './HttpServer'
-export * from './restful/v1'
 export * from './common/handler'

core/src/node/api/processors/app.test.ts

@@ -2,7 +2,6 @@ jest.mock('../../helper', () => ({
   ...jest.requireActual('../../helper'),
   getJanDataFolderPath: () => './app',
 }))
-import { dirname } from 'path'
 import { App } from './app'
 
 it('should call stopServer', () => {

core/src/node/api/processors/app.ts

@@ -3,7 +3,6 @@ import { basename, dirname, isAbsolute, join, relative } from 'path'
 import { Processor } from './Processor'
 import {
   log as writeLog,
-  appResourcePath,
   getAppConfigurations as appConfiguration,
   updateAppConfiguration,
   normalizeFilePath,
@@ -91,8 +90,6 @@ export class App implements Processor {
       port: args?.port,
       isCorsEnabled: args?.isCorsEnabled,
       isVerboseEnabled: args?.isVerboseEnabled,
-      schemaPath: join(appResourcePath(), 'docs', 'openapi', 'jan.yaml'),
-      baseDir: join(appResourcePath(), 'docs', 'openapi'),
       prefix: args?.prefix,
     })
   }

core/src/node/api/restful/app/download.test.ts

@@ -1,62 +0,0 @@
-import { HttpServer } from '../../HttpServer'
-import { DownloadManager } from '../../../helper/download'
-
-describe('downloadRouter', () => {
-  let app: HttpServer
-
-  beforeEach(() => {
-    app = {
-      register: jest.fn(),
-      post: jest.fn(),
-      get: jest.fn(),
-      patch: jest.fn(),
-      put: jest.fn(),
-      delete: jest.fn(),
-    }
-  })
-
-  it('should return download progress for a given modelId', async () => {
-    const modelId = '123'
-    const downloadProgress = { progress: 50 }
-
-    DownloadManager.instance.downloadProgressMap[modelId] = downloadProgress as any
-
-    const req = { params: { modelId } }
-    const res = {
-      status: jest.fn(),
-      send: jest.fn(),
-    }
-
-    jest.spyOn(app, 'get').mockImplementation((path, handler) => {
-      if (path === `/download/getDownloadProgress/${modelId}`) {
-        res.status(200)
-        res.send(downloadProgress)
-      }
-    })
-
-    app.get(`/download/getDownloadProgress/${modelId}`, req as any)
-    expect(res.status).toHaveBeenCalledWith(200)
-    expect(res.send).toHaveBeenCalledWith(downloadProgress)
-  })
-
-  it('should return 404 if download progress is not found', async () => {
-    const modelId = '123'
-
-    const req = { params: { modelId } }
-    const res = {
-      status: jest.fn(),
-      send: jest.fn(),
-    }
-
-
-    jest.spyOn(app, 'get').mockImplementation((path, handler) => {
-      if (path === `/download/getDownloadProgress/${modelId}`) {
-        res.status(404)
-        res.send({ message: 'Download progress not found' })
-      }
-    })
-    app.get(`/download/getDownloadProgress/${modelId}`, req as any)
-    expect(res.status).toHaveBeenCalledWith(404)
-    expect(res.send).toHaveBeenCalledWith({ message: 'Download progress not found' })
-  })
-})

core/src/node/api/restful/app/download.ts

@@ -1,23 +0,0 @@
-import { DownloadRoute } from '../../../../types/api'
-import { DownloadManager } from '../../../helper/download'
-import { HttpServer } from '../../HttpServer'
-
-export const downloadRouter = async (app: HttpServer) => {
-  app.get(`/download/${DownloadRoute.getDownloadProgress}/:modelId`, async (req, res) => {
-    const modelId = req.params.modelId
-
-    console.debug(`Getting download progress for model ${modelId}`)
-    console.debug(
-      `All Download progress: ${JSON.stringify(DownloadManager.instance.downloadProgressMap)}`
-    )
-
-    // check if null DownloadManager.instance.downloadProgressMap
-    if (!DownloadManager.instance.downloadProgressMap[modelId]) {
-      return res.status(404).send({
-        message: 'Download progress not found',
-      })
-    } else {
-      return res.status(200).send(DownloadManager.instance.downloadProgressMap[modelId])
-    }
-  })
-}

core/src/node/api/restful/app/handlers.test.ts

@@ -1,16 +0,0 @@
-// 
-import { jest } from '@jest/globals';
-
-import { HttpServer } from '../../HttpServer';
-import { handleRequests } from './handlers';
-import { Handler, RequestHandler } from '../../common/handler';
-
-it('should initialize RequestHandler and call handle', () => {
-  const mockHandle = jest.fn();
-  jest.spyOn(RequestHandler.prototype, 'handle').mockImplementation(mockHandle);
-  
-  const mockApp = { post: jest.fn() };
-  handleRequests(mockApp as unknown as HttpServer);
-  
-  expect(mockHandle).toHaveBeenCalled();
-});

core/src/node/api/restful/app/handlers.ts

@@ -1,13 +0,0 @@
-import { HttpServer } from '../../HttpServer'
-import { Handler, RequestHandler } from '../../common/handler'
-
-export function handleRequests(app: HttpServer) {
-  const restWrapper: Handler = (route: string, listener: (...args: any[]) => any) => {
-    app.post(`/app/${route}`, async (request: any, reply: any) => {
-      const args = JSON.parse(request.body) as any[]
-      reply.send(JSON.stringify(await listener(...args)))
-    })
-  }
-  const handler = new RequestHandler(restWrapper)
-  handler.handle()
-}

core/src/node/api/restful/common.test.ts

@@ -1,21 +0,0 @@
-
-import { commonRouter } from './common';
-import { JanApiRouteConfiguration } from './helper/configuration';
-
-test('commonRouter sets up routes for each key in JanApiRouteConfiguration', async () => {
-  const mockHttpServer = {
-    get: jest.fn(),
-    post: jest.fn(),
-    patch: jest.fn(),
-    put: jest.fn(),
-    delete: jest.fn(),
-  };
-  await commonRouter(mockHttpServer as any);
-
-  const expectedRoutes = Object.keys(JanApiRouteConfiguration);
-  expectedRoutes.forEach((key) => {
-    expect(mockHttpServer.get).toHaveBeenCalledWith(`/${key}`, expect.any(Function));
-    expect(mockHttpServer.get).toHaveBeenCalledWith(`/${key}/:id`, expect.any(Function));
-    expect(mockHttpServer.delete).toHaveBeenCalledWith(`/${key}/:id`, expect.any(Function));
-  });
-});

core/src/node/api/restful/common.ts

@@ -1,82 +0,0 @@
-import { HttpServer } from '../HttpServer'
-import {
-  chatCompletions,
-  downloadModel,
-  getBuilder,
-  retrieveBuilder,
-  createMessage,
-  createThread,
-  getMessages,
-  retrieveMessage,
-  updateThread,
-  models,
-} from './helper/builder'
-
-import { JanApiRouteConfiguration } from './helper/configuration'
-
-export const commonRouter = async (app: HttpServer) => {
-  const normalizeData = (data: any) => {
-    return {
-      object: 'list',
-      data,
-    }
-  }
-  // Common Routes
-  // Read & Delete :: Threads | Models | Assistants
-  Object.keys(JanApiRouteConfiguration).forEach((key) => {
-    app.get(`/${key}`, async (_req, _res) => {
-      if (key.includes('models')) {
-        return models(_req, _res)
-      }
-      return getBuilder(JanApiRouteConfiguration[key]).then(normalizeData)
-    })
-
-    app.get(`/${key}/:id`, async (_req: any, _res: any) => {
-      if (key.includes('models')) {
-        return models(_req, _res)
-      }
-      return retrieveBuilder(JanApiRouteConfiguration[key], _req.params.id)
-    })
-
-    app.delete(`/${key}/:id`, async (_req: any, _res: any) => {
-      if (key.includes('models')) {
-        return models(_req, _res)
-      }
-      return retrieveBuilder(JanApiRouteConfiguration[key], _req.params.id)
-    })
-  })
-
-  // Threads
-  app.post(`/threads`, async (req, res) => createThread(req.body))
-
-  app.get(`/threads/:threadId/messages`, async (req, res) =>
-    getMessages(req.params.threadId).then(normalizeData)
-  )
-
-  app.get(`/threads/:threadId/messages/:messageId`, async (req, res) =>
-    retrieveMessage(req.params.threadId, req.params.messageId)
-  )
-
-  app.post(`/threads/:threadId/messages`, async (req, res) =>
-    createMessage(req.params.threadId as any, req.body as any)
-  )
-
-  app.patch(`/threads/:threadId`, async (request: any) =>
-    updateThread(request.params.threadId, request.body)
-  )
-
-  // Models
-  app.get(`/models/download/:modelId`, async (request: any) =>
-    downloadModel(request.params.modelId, {
-      ignoreSSL: request.query.ignoreSSL === 'true',
-      proxy: request.query.proxy,
-    })
-  )
-
-  app.post(`/models/start`, async (request: any, reply: any) => models(request, reply))
-
-  app.post(`/models/stop`, async (request: any, reply: any) => models(request, reply))
-
-  // Chat Completion
-  app.post(`/chat/completions`, async (request: any, reply: any) => chatCompletions(request, reply))
-}

core/src/node/api/restful/helper/builder.test.ts

@@ -1,251 +0,0 @@
-import { existsSync, readdirSync, readFileSync, writeFileSync, mkdirSync, appendFileSync } from 'fs'
-import {
-  getBuilder,
-  retrieveBuilder,
-  getMessages,
-  retrieveMessage,
-  createThread,
-  updateThread,
-  createMessage,
-  downloadModel,
-  chatCompletions,
-} from './builder'
-import { RouteConfiguration } from './configuration'
-
-jest.mock('fs')
-jest.mock('path')
-jest.mock('../../../helper', () => ({
-  getEngineConfiguration: jest.fn(),
-  getJanDataFolderPath: jest.fn().mockReturnValue('/mock/path'),
-}))
-jest.mock('request')
-jest.mock('request-progress')
-jest.mock('node-fetch')
-
-describe('builder helper functions', () => {
-  const mockConfiguration: RouteConfiguration = {
-    dirName: 'mockDir',
-    metadataFileName: 'metadata.json',
-    delete: {
-      object: 'mockObject',
-    },
-  }
-
-  beforeEach(() => {
-    jest.clearAllMocks()
-  })
-
-  describe('getBuilder', () => {
-    it('should return an empty array if directory does not exist', async () => {
-      ;(existsSync as jest.Mock).mockReturnValue(false)
-      const result = await getBuilder(mockConfiguration)
-      expect(result).toEqual([])
-    })
-
-    it('should return model data if directory exists', async () => {
-      ;(existsSync as jest.Mock).mockReturnValue(true)
-      ;(readdirSync as jest.Mock).mockReturnValue(['file1'])
-      ;(readFileSync as jest.Mock).mockReturnValue(JSON.stringify({ id: 'model1' }))
-
-      const result = await getBuilder(mockConfiguration)
-      expect(result).toEqual([{ id: 'model1' }])
-    })
-  })
-
-  describe('retrieveBuilder', () => {
-    it('should return undefined if no data matches the id', async () => {
-      ;(existsSync as jest.Mock).mockReturnValue(true)
-      ;(readdirSync as jest.Mock).mockReturnValue(['file1'])
-      ;(readFileSync as jest.Mock).mockReturnValue(JSON.stringify({ id: 'model1' }))
-
-      const result = await retrieveBuilder(mockConfiguration, 'nonexistentId')
-      expect(result).toBeUndefined()
-    })
-
-    it('should return the matching data', async () => {
-      ;(existsSync as jest.Mock).mockReturnValue(true)
-      ;(readdirSync as jest.Mock).mockReturnValue(['file1'])
-      ;(readFileSync as jest.Mock).mockReturnValue(JSON.stringify({ id: 'model1' }))
-
-      const result = await retrieveBuilder(mockConfiguration, 'model1')
-      expect(result).toEqual({ id: 'model1' })
-    })
-  })
-
-  describe('getMessages', () => {
-    it('should return an empty array if message file does not exist', async () => {
-      ;(existsSync as jest.Mock).mockReturnValue(false)
-
-      const result = await getMessages('thread1')
-      expect(result).toEqual([])
-    })
-
-    it('should return messages if message file exists', async () => {
-      ;(existsSync as jest.Mock).mockReturnValue(true)
-      ;(readdirSync as jest.Mock).mockReturnValue(['messages.jsonl'])
-      ;(readFileSync as jest.Mock).mockReturnValue('{"id":"msg1"}\n{"id":"msg2"}\n')
-
-      const result = await getMessages('thread1')
-      expect(result).toEqual([{ id: 'msg1' }, { id: 'msg2' }])
-    })
-  })
-
-  describe('retrieveMessage', () => {
-    it('should return a message if no messages match the id', async () => {
-      ;(existsSync as jest.Mock).mockReturnValue(true)
-      ;(readdirSync as jest.Mock).mockReturnValue(['messages.jsonl'])
-      ;(readFileSync as jest.Mock).mockReturnValue('{"id":"msg1"}\n')
-
-      const result = await retrieveMessage('thread1', 'nonexistentId')
-      expect(result).toEqual({ message: 'Not found' })
-    })
-
-    it('should return the matching message', async () => {
-      ;(existsSync as jest.Mock).mockReturnValue(true)
-      ;(readdirSync as jest.Mock).mockReturnValue(['messages.jsonl'])
-      ;(readFileSync as jest.Mock).mockReturnValue('{"id":"msg1"}\n')
-
-      const result = await retrieveMessage('thread1', 'msg1')
-      expect(result).toEqual({ id: 'msg1' })
-    })
-  })
-
-  describe('createThread', () => {
-    it('should return a message if thread has no assistants', async () => {
-      const result = await createThread({})
-      expect(result).toEqual({ message: 'Thread must have at least one assistant' })
-    })
-
-    it('should create a thread and return the updated thread', async () => {
-      ;(existsSync as jest.Mock).mockReturnValue(false)
-
-      const thread = { assistants: [{ assistant_id: 'assistant1' }] }
-      const result = await createThread(thread)
-      expect(mkdirSync).toHaveBeenCalled()
-      expect(writeFileSync).toHaveBeenCalled()
-      expect(result.id).toBeDefined()
-    })
-  })
-
-  describe('updateThread', () => {
-    it('should return a message if thread is not found', async () => {
-      ;(existsSync as jest.Mock).mockReturnValue(true)
-      ;(readdirSync as jest.Mock).mockReturnValue(['file1'])
-      ;(readFileSync as jest.Mock).mockReturnValue(JSON.stringify({ id: 'model1' }))
-
-      const result = await updateThread('nonexistentId', {})
-      expect(result).toEqual({ message: 'Thread not found' })
-    })
-
-    it('should update the thread and return the updated thread', async () => {
-      ;(existsSync as jest.Mock).mockReturnValue(true)
-      ;(readdirSync as jest.Mock).mockReturnValue(['file1'])
-      ;(readFileSync as jest.Mock).mockReturnValue(JSON.stringify({ id: 'model1' }))
-
-      const result = await updateThread('model1', { name: 'updatedName' })
-      expect(writeFileSync).toHaveBeenCalled()
-      expect(result.name).toEqual('updatedName')
-    })
-  })
-
-  describe('createMessage', () => {
-    it('should create a message and return the created message', async () => {
-      ;(existsSync as jest.Mock).mockReturnValue(false)
-      const message = { role: 'user', content: 'Hello' }
-
-      const result = (await createMessage('thread1', message)) as any
-      expect(mkdirSync).toHaveBeenCalled()
-      expect(appendFileSync).toHaveBeenCalled()
-      expect(result.id).toBeDefined()
-    })
-  })
-
-  describe('downloadModel', () => {
-    it('should return a message if model is not found', async () => {
-      ;(existsSync as jest.Mock).mockReturnValue(true)
-      ;(readdirSync as jest.Mock).mockReturnValue(['file1'])
-      ;(readFileSync as jest.Mock).mockReturnValue(JSON.stringify({ id: 'model1' }))
-
-      const result = await downloadModel('nonexistentId')
-      expect(result).toEqual({ message: 'Model not found' })
-    })
-
-    it('should start downloading the model', async () => {
-      ;(existsSync as jest.Mock).mockReturnValue(true)
-      ;(readdirSync as jest.Mock).mockReturnValue(['file1'])
-      ;(readFileSync as jest.Mock).mockReturnValue(
-        JSON.stringify({ id: 'model1', object: 'model', sources: ['http://example.com'] })
-      )
-      const result = await downloadModel('model1')
-      expect(result).toEqual({ message: 'Starting download model1' })
-    })
-  })
-
-  describe('chatCompletions', () => {
-    it('should return the error on status not ok', async () => {
-      const request = { body: { model: 'model1' } }
-      const mockSend = jest.fn()
-      const reply = {
-        code: jest.fn().mockReturnThis(),
-        send: jest.fn(),
-        headers: jest.fn().mockReturnValue({
-          send: mockSend,
-        }),
-        raw: {
-          writeHead: jest.fn(),
-          pipe: jest.fn(),
-        },
-      }
-
-      ;(existsSync as jest.Mock).mockReturnValue(true)
-      ;(readdirSync as jest.Mock).mockReturnValue(['file1'])
-      ;(readFileSync as jest.Mock).mockReturnValue(
-        JSON.stringify({ id: 'model1', engine: 'openai' })
-      )
-
-      // Mock fetch
-      const fetch = require('node-fetch')
-      fetch.mockResolvedValue({
-        status: 400,
-        headers: new Map([
-          ['content-type', 'application/json'],
-          ['x-request-id', '123456'],
-        ]),
-        body: { pipe: jest.fn() },
-        text: jest.fn().mockResolvedValue({ error: 'Mock error response' }),
-      })
-      await chatCompletions(request, reply)
-      expect(reply.code).toHaveBeenCalledWith(400)
-      expect(mockSend).toHaveBeenCalledWith(
-        expect.objectContaining({
-          error: 'Mock error response',
-        })
-      )
-    })
-
-    it('should return the chat completions', async () => {
-      const request = { body: { model: 'model1' } }
-      const reply = {
-        code: jest.fn().mockReturnThis(),
-        send: jest.fn(),
-        raw: { writeHead: jest.fn(), pipe: jest.fn() },
-      }
-
-      ;(existsSync as jest.Mock).mockReturnValue(true)
-      ;(readdirSync as jest.Mock).mockReturnValue(['file1'])
-      ;(readFileSync as jest.Mock).mockReturnValue(
-        JSON.stringify({ id: 'model1', engine: 'openai' })
-      )
-
-      // Mock fetch
-      const fetch = require('node-fetch')
-      fetch.mockResolvedValue({
-        status: 200,
-        body: { pipe: jest.fn() },
-        json: jest.fn().mockResolvedValue({ completions: ['completion1'] }),
-      })
-      await chatCompletions(request, reply)
-      expect(reply.raw.writeHead).toHaveBeenCalledWith(200, expect.any(Object))
-    })
-  })
-})

core/src/node/api/restful/helper/builder.ts

@@ -1,340 +0,0 @@
-import {
-  existsSync,
-  readdirSync,
-  readFileSync,
-  writeFileSync,
-  mkdirSync,
-  appendFileSync,
-  createWriteStream,
-  rmdirSync,
-} from 'fs'
-import { JanApiRouteConfiguration, RouteConfiguration } from './configuration'
-import { join } from 'path'
-import { ContentType, InferenceEngine, MessageStatus, ThreadMessage } from '../../../../types'
-import { getJanDataFolderPath } from '../../../helper'
-import { CORTEX_API_URL } from './consts'
-
-// TODO: Refactor these
-export const getBuilder = async (configuration: RouteConfiguration) => {
-  const directoryPath = join(getJanDataFolderPath(), configuration.dirName)
-  try {
-    if (!existsSync(directoryPath)) {
-      console.debug('model folder not found')
-      return []
-    }
-
-    const files: string[] = readdirSync(directoryPath)
-
-    const allDirectories: string[] = []
-    for (const file of files) {
-      if (file === '.DS_Store') continue
-      allDirectories.push(file)
-    }
-
-    const results = allDirectories
-      .map((dirName) => {
-        const jsonPath = join(directoryPath, dirName, configuration.metadataFileName)
-        return readModelMetadata(jsonPath)
-      })
-      .filter((data) => !!data)
-    const modelData = results
-      .map((result: any) => {
-        try {
-          return JSON.parse(result)
-        } catch (err) {
-          console.error(err)
-        }
-      })
-      .filter((e: any) => !!e)
-
-    return modelData
-  } catch (err) {
-    console.error(err)
-    return []
-  }
-}
-
-const readModelMetadata = (path: string): string | undefined => {
-  if (existsSync(path)) {
-    return readFileSync(path, 'utf-8')
-  } else {
-    return undefined
-  }
-}
-
-export const retrieveBuilder = async (configuration: RouteConfiguration, id: string) => {
-  const data = await getBuilder(configuration)
-  const filteredData = data.filter((d: any) => d.id === id)[0]
-
-  if (!filteredData) {
-    return undefined
-  }
-
-  return filteredData
-}
-
-export const getMessages = async (threadId: string): Promise<ThreadMessage[]> => {
-  const threadDirPath = join(getJanDataFolderPath(), 'threads', threadId)
-  const messageFile = 'messages.jsonl'
-  try {
-    const files: string[] = readdirSync(threadDirPath)
-    if (!files.includes(messageFile)) {
-      console.error(`${threadDirPath} not contains message file`)
-      return []
-    }
-
-    const messageFilePath = join(threadDirPath, messageFile)
-    if (!existsSync(messageFilePath)) {
-      console.debug('message file not found')
-      return []
-    }
-
-    const lines = readFileSync(messageFilePath, 'utf-8')
-      .toString()
-      .split('\n')
-      .filter((line: any) => line !== '')
-
-    const messages: ThreadMessage[] = []
-    lines.forEach((line: string) => {
-      messages.push(JSON.parse(line) as ThreadMessage)
-    })
-    return messages
-  } catch (err) {
-    console.error(err)
-    return []
-  }
-}
-
-export const retrieveMessage = async (threadId: string, messageId: string) => {
-  const messages = await getMessages(threadId)
-  const filteredMessages = messages.filter((m) => m.id === messageId)
-  if (!filteredMessages || filteredMessages.length === 0) {
-    return {
-      message: 'Not found',
-    }
-  }
-
-  return filteredMessages[0]
-}
-
-export const createThread = async (thread: any) => {
-  const threadMetadataFileName = 'thread.json'
-  // TODO: add validation
-  if (!thread.assistants || thread.assistants.length === 0) {
-    return {
-      message: 'Thread must have at least one assistant',
-    }
-  }
-
-  const threadId = generateThreadId(thread.assistants[0].assistant_id)
-  try {
-    const updatedThread = {
-      ...thread,
-      id: threadId,
-      created: Date.now(),
-      updated: Date.now(),
-    }
-    const threadDirPath = join(getJanDataFolderPath(), 'threads', updatedThread.id)
-    const threadJsonPath = join(threadDirPath, threadMetadataFileName)
-
-    if (!existsSync(threadDirPath)) {
-      mkdirSync(threadDirPath)
-    }
-
-    await writeFileSync(threadJsonPath, JSON.stringify(updatedThread, null, 2))
-    return updatedThread
-  } catch (err) {
-    return {
-      error: err,
-    }
-  }
-}
-
-export const updateThread = async (threadId: string, thread: any) => {
-  const threadMetadataFileName = 'thread.json'
-  const currentThreadData = await retrieveBuilder(JanApiRouteConfiguration.threads, threadId)
-  if (!currentThreadData) {
-    return {
-      message: 'Thread not found',
-    }
-  }
-  // we don't want to update the id and object
-  delete thread.id
-  delete thread.object
-
-  const updatedThread = {
-    ...currentThreadData,
-    ...thread,
-    updated: Date.now(),
-  }
-  try {
-    const threadDirPath = join(getJanDataFolderPath(), 'threads', updatedThread.id)
-    const threadJsonPath = join(threadDirPath, threadMetadataFileName)
-
-    await writeFileSync(threadJsonPath, JSON.stringify(updatedThread, null, 2))
-    return updatedThread
-  } catch (err) {
-    return {
-      message: err,
-    }
-  }
-}
-
-const generateThreadId = (assistantId: string) => {
-  return `${assistantId}_${(Date.now() / 1000).toFixed(0)}`
-}
-
-export const createMessage = async (threadId: string, message: any) => {
-  const threadMessagesFileName = 'messages.jsonl'
-
-  try {
-    const { ulid } = require('ulidx')
-    const msgId = ulid()
-    const createdAt = Date.now()
-    const threadMessage: ThreadMessage = {
-      id: msgId,
-      thread_id: threadId,
-      status: MessageStatus.Ready,
-      created: createdAt,
-      updated: createdAt,
-      object: 'thread.message',
-      role: message.role,
-      content: [
-        {
-          type: ContentType.Text,
-          text: {
-            value: message.content,
-            annotations: [],
-          },
-        },
-      ],
-    }
-
-    const threadDirPath = join(getJanDataFolderPath(), 'threads', threadId)
-    const threadMessagePath = join(threadDirPath, threadMessagesFileName)
-
-    if (!existsSync(threadDirPath)) {
-      mkdirSync(threadDirPath)
-    }
-    appendFileSync(threadMessagePath, JSON.stringify(threadMessage) + '\n')
-    return threadMessage
-  } catch (err) {
-    return {
-      message: err,
-    }
-  }
-}
-
-export const downloadModel = async (
-  modelId: string,
-  network?: { proxy?: string; ignoreSSL?: boolean }
-) => {
-  const strictSSL = !network?.ignoreSSL
-  const proxy = network?.proxy?.startsWith('http') ? network.proxy : undefined
-  const model = await retrieveBuilder(JanApiRouteConfiguration.models, modelId)
-  if (!model || model.object !== 'model') {
-    return {
-      message: 'Model not found',
-    }
-  }
-
-  const directoryPath = join(getJanDataFolderPath(), 'models', modelId)
-  if (!existsSync(directoryPath)) {
-    mkdirSync(directoryPath)
-  }
-
-  // path to model binary
-  const modelBinaryPath = join(directoryPath, modelId)
-
-  const request = require('request')
-  const progress = require('request-progress')
-
-  for (const source of model.sources) {
-    const rq = request({ url: source, strictSSL, proxy })
-    progress(rq, {})
-      ?.on('progress', function (state: any) {
-        console.debug('progress', JSON.stringify(state, null, 2))
-      })
-      ?.on('error', function (err: Error) {
-        console.error('error', err)
-      })
-      ?.on('end', function () {
-        console.debug('end')
-      })
-      .pipe(createWriteStream(modelBinaryPath))
-  }
-
-  return {
-    message: `Starting download ${modelId}`,
-  }
-}
-
-/**
- * Proxy /models to cortex
- * @param request
- * @param reply
- */
-export const models = async (request: any, reply: any) => {
-  const fetch = require('node-fetch')
-  const headers: Record<string, any> = {
-    'Content-Type': 'application/json',
-  }
-
-  const response = await fetch(`${CORTEX_API_URL}/models${request.url.split('/models')[1] ?? ""}`, {
-    method: request.method,
-    headers: headers,
-    body: JSON.stringify(request.body),
-  })
-
-  if (response.status !== 200) {
-    // Forward the error response to client via reply
-    const responseBody = await response.text()
-    const responseHeaders = Object.fromEntries(response.headers)
-    reply.code(response.status).headers(responseHeaders).send(responseBody)
-  } else {
-    reply.raw.writeHead(200, {
-      'Content-Type': 'application/json',
-      'Cache-Control': 'no-cache',
-      'Connection': 'keep-alive',
-      'Access-Control-Allow-Origin': '*',
-    })
-    response.body.pipe(reply.raw)
-  }
-}
-
-/**
- * Proxy chat completions
- * @param request
- * @param reply
- */
-export const chatCompletions = async (request: any, reply: any) => {
-  const headers: Record<string, any> = {
-    'Content-Type': 'application/json',
-  }
-
-  // add engine for new cortex cpp engine
-  if (request.body.engine === InferenceEngine.nitro) {
-    request.body.engine = InferenceEngine.cortex_llamacpp
-  }
-
-  const fetch = require('node-fetch')
-  const response = await fetch(`${CORTEX_API_URL}/chat/completions`, {
-    method: 'POST',
-    headers: headers,
-    body: JSON.stringify(request.body),
-  })
-  if (response.status !== 200) {
-    // Forward the error response to client via reply
-    const responseBody = await response.text()
-    const responseHeaders = Object.fromEntries(response.headers)
-    reply.code(response.status).headers(responseHeaders).send(responseBody)
-  } else {
-    reply.raw.writeHead(200, {
-      'Content-Type': request.body.stream === true ? 'text/event-stream' : 'application/json',
-      'Cache-Control': 'no-cache',
-      'Connection': 'keep-alive',
-      'Access-Control-Allow-Origin': '*',
-    })
-    response.body.pipe(reply.raw)
-  }
-}

core/src/node/api/restful/helper/configuration.test.ts

@@ -1,24 +0,0 @@
-import { JanApiRouteConfiguration } from './configuration'
-
-describe('JanApiRouteConfiguration', () => {
-  it('should have the correct models configuration', () => {
-    const modelsConfig = JanApiRouteConfiguration.models;
-    expect(modelsConfig.dirName).toBe('models');
-    expect(modelsConfig.metadataFileName).toBe('model.json');
-    expect(modelsConfig.delete.object).toBe('model');
-  });
-
-  it('should have the correct assistants configuration', () => {
-    const assistantsConfig = JanApiRouteConfiguration.assistants;
-    expect(assistantsConfig.dirName).toBe('assistants');
-    expect(assistantsConfig.metadataFileName).toBe('assistant.json');
-    expect(assistantsConfig.delete.object).toBe('assistant');
-  });
-
-  it('should have the correct threads configuration', () => {
-    const threadsConfig = JanApiRouteConfiguration.threads;
-    expect(threadsConfig.dirName).toBe('threads');
-    expect(threadsConfig.metadataFileName).toBe('thread.json');
-    expect(threadsConfig.delete.object).toBe('thread');
-  });
-});

core/src/node/api/restful/helper/configuration.ts

@@ -1,31 +0,0 @@
-export const JanApiRouteConfiguration: Record<string, RouteConfiguration> = {
-  models: {
-    dirName: 'models',
-    metadataFileName: 'model.json',
-    delete: {
-      object: 'model',
-    },
-  },
-  assistants: {
-    dirName: 'assistants',
-    metadataFileName: 'assistant.json',
-    delete: {
-      object: 'assistant',
-    },
-  },
-  threads: {
-    dirName: 'threads',
-    metadataFileName: 'thread.json',
-    delete: {
-      object: 'thread',
-    },
-  },
-}
-
-export type RouteConfiguration = {
-  dirName: string
-  metadataFileName: string
-  delete: {
-    object: string
-  }
-}

core/src/node/api/restful/helper/consts.test.ts

@@ -1,5 +0,0 @@
-import { CORTEX_DEFAULT_PORT } from './consts'
-
-it('should test CORTEX_DEFAULT_PORT', () => {
-  expect(CORTEX_DEFAULT_PORT).toBe(39291)
-})

core/src/node/api/restful/helper/consts.ts

@@ -1,7 +0,0 @@
-export const CORTEX_DEFAULT_PORT = 39291
-
-export const LOCAL_HOST = '127.0.0.1'
-
-export const SUPPORTED_MODEL_FORMAT = '.gguf'
-
-export const CORTEX_API_URL = `http://${LOCAL_HOST}:${CORTEX_DEFAULT_PORT}/v1`

core/src/node/api/restful/v1.test.ts

@@ -1,16 +0,0 @@
-
-import { v1Router } from './v1';
-import { commonRouter } from './common';
-
-test('should define v1Router function', () => {
-  expect(v1Router).toBeDefined();
-});
-
-test('should register commonRouter', () => {
-  const mockApp = {
-    register: jest.fn(),
-  };
-  v1Router(mockApp as any);
-  expect(mockApp.register).toHaveBeenCalledWith(commonRouter);
-});
-

core/src/node/api/restful/v1.ts

@@ -1,16 +0,0 @@
-import { HttpServer } from '../HttpServer'
-import { commonRouter } from './common'
-
-export const v1Router = async (app: HttpServer) => {
-  // MARK: Public API Routes
-  app.register(commonRouter)
-
-  // MARK: Internal Application Routes
-  // DEPRECATED: Vulnerability possible issues
-  // handleRequests(app)
-
-  // Expanded route for tracking download progress
-  // TODO: Replace by Observer Wrapper (ZeroMQ / Vanilla Websocket)
-  // DEPRECATED: Jan FE Docker deploy is deprecated
-  // app.register(downloadRouter)
-}

core/src/node/helper/path.ts

@@ -1,5 +1,4 @@
-import { join, resolve } from 'path'
-import { getJanDataFolderPath } from './config'
+import { join } from 'path'
 
 /**
  * Normalize file path
@@ -34,4 +33,5 @@ export function appResourcePath() {
 
   // server
   return join(global.core.appPath(), '../../..')
-}
+}
+

core/src/types/assistant/assistantEntity.ts

@@ -36,3 +36,10 @@ export type Assistant = {
   /** Represents the metadata of the object. */
   metadata?: Record<string, unknown>
 }
+
+export interface CodeInterpreterTool {
+  /**
+   * The type of tool being defined: `code_interpreter`
+   */
+  type: 'code_interpreter'
+}

core/src/types/message/messageEntity.ts

@@ -1,3 +1,4 @@
+import { CodeInterpreterTool } from '../assistant'
 import { ChatCompletionMessage, ChatCompletionRole } from '../inference'
 import { ModelInfo } from '../model'
 import { Thread } from '../thread'
@@ -15,6 +16,10 @@ export type ThreadMessage = {
   thread_id: string
   /** The assistant id of this thread. **/
   assistant_id?: string
+  /**
+   * A list of files attached to the message, and the tools they were added to.
+   */
+  attachments?: Array<Attachment> | null
   /** The role of the author of this message. **/
   role: ChatCompletionRole
   /** The content of this message. **/
@@ -22,9 +27,9 @@ export type ThreadMessage = {
   /** The status of this message. **/
   status: MessageStatus
   /** The timestamp indicating when this message was created. Represented in Unix time. **/
-  created: number
+  created_at: number
   /** The timestamp indicating when this message was updated. Represented in Unix time. **/
-  updated: number
+  completed_at: number
   /** The additional metadata of this message. **/
   metadata?: Record<string, unknown>
 
@@ -52,6 +57,11 @@ export type MessageRequest = {
    */
   assistantId?: string
 
+  /**
+   * A list of files attached to the message, and the tools they were added to.
+   */
+  attachments: Array<Attachment> | null
+
   /** Messages for constructing a chat completion request **/
   messages?: ChatCompletionMessage[]
 
@@ -97,8 +107,7 @@ export enum ErrorCode {
  */
 export enum ContentType {
   Text = 'text',
-  Image = 'image',
-  Pdf = 'pdf',
+  Image = 'image_url',
 }
 
 /**
@@ -108,8 +117,15 @@ export enum ContentType {
 export type ContentValue = {
   value: string
   annotations: string[]
-  name?: string
-  size?: number
+}
+
+/**
+ * The `ImageContentValue` type defines the shape of a content value object of image type
+ * @data_transfer_object
+ */
+export type ImageContentValue = {
+  detail?: string
+  url?: string
 }
 
 /**
@@ -118,5 +134,37 @@ export type ContentValue = {
  */
 export type ThreadContent = {
   type: ContentType
-  text: ContentValue
+  text?: ContentValue
+  image_url?: ImageContentValue
+}
+
+export interface Attachment {
+  /**
+   * The ID of the file to attach to the message.
+   */
+  file_id?: string
+
+  /**
+   * The tools to add this file to.
+   */
+  tools?: Array<CodeInterpreterTool | Attachment.AssistantToolsFileSearchTypeOnly>
+}
+
+export namespace Attachment {
+  export interface AssistantToolsFileSearchTypeOnly {
+    /**
+     * The type of tool being defined: `file_search`
+     */
+    type: 'file_search'
+  }
+}
+
+/**
+ * On an incomplete message, details about why the message is incomplete.
+ */
+export interface IncompleteDetails {
+  /**
+   * The reason the message is incomplete.
+   */
+  reason: 'content_filter' | 'max_tokens' | 'run_cancelled' | 'run_expired' | 'run_failed'
 }

core/src/types/message/messageInterface.ts

@@ -11,20 +11,20 @@ export interface MessageInterface {
    * @param {ThreadMessage} message - The message to be added.
    * @returns {Promise<void>} A promise that resolves when the message has been added.
    */
-  addNewMessage(message: ThreadMessage): Promise<void>
-
-  /**
-   * Writes an array of messages to a specific thread.
-   * @param {string} threadId - The ID of the thread to write the messages to.
-   * @param {ThreadMessage[]} messages - The array of messages to be written.
-   * @returns {Promise<void>} A promise that resolves when the messages have been written.
-   */
-  writeMessages(threadId: string, messages: ThreadMessage[]): Promise<void>
+  createMessage(message: ThreadMessage): Promise<ThreadMessage>
 
   /**
    * Retrieves all messages from a specific thread.
    * @param {string} threadId - The ID of the thread to retrieve the messages from.
    * @returns {Promise<ThreadMessage[]>} A promise that resolves to an array of messages from the thread.
    */
-  getAllMessages(threadId: string): Promise<ThreadMessage[]>
+  listMessages(threadId: string): Promise<ThreadMessage[]>
+
+  /**
+   * Deletes a specific message from a thread.
+   * @param {string} threadId - The ID of the thread from which the message will be deleted.
+   * @param {string} messageId - The ID of the message to be deleted.
+   * @returns {Promise<void>} A promise that resolves when the message has been successfully deleted.
+   */
+  deleteMessage(threadId: string, messageId: string): Promise<void>
 }

core/src/types/thread/threadInterface.ts

@@ -11,15 +11,23 @@ export interface ThreadInterface {
    * @abstract
    * @returns {Promise<Thread[]>} A promise that resolves to an array of threads.
    */
-  getThreads(): Promise<Thread[]>
+  listThreads(): Promise<Thread[]>
 
   /**
-   * Saves a thread.
+   * Create a thread.
    * @abstract
    * @param {Thread} thread - The thread to save.
    * @returns {Promise<void>} A promise that resolves when the thread is saved.
    */
-  saveThread(thread: Thread): Promise<void>
+  createThread(thread: Thread): Promise<Thread>
+
+  /**
+   * modify a thread.
+   * @abstract
+   * @param {Thread} thread - The thread to save.
+   * @returns {Promise<void>} A promise that resolves when the thread is saved.
+   */
+  modifyThread(thread: Thread): Promise<void>
 
   /**
    * Deletes a thread.

core/tsconfig.json

@@ -13,7 +13,8 @@
     "declarationDir": "dist/types",
     "outDir": "dist/lib",
     "importHelpers": true,
-    "types": ["@types/jest"]
+    "types": ["@types/jest"],
+    "resolveJsonModule": true
   },
   "include": ["src"],
   "exclude": ["**/*.test.ts"]

docs/openapi/specs/assistants.yaml

@@ -1,319 +0,0 @@
----
-components:
-  schemas:
-    AssistantObject:
-      type: object
-      properties:
-        id:
-          type: string
-          description: The identifier of the assistant.
-          example: asst_abc123
-        object:
-          type: string
-          description: Type of the object, indicating it's an assistant.
-          default: assistant
-        version:
-          type: integer
-          description: Version number of the assistant.
-          example: 1
-        created_at:
-          type: integer
-          format: int64
-          description: Unix timestamp representing the creation time of the assistant.
-          example: 1698984975
-        name:
-          type: string
-          description: Name of the assistant.
-          example: Math Tutor
-        description:
-          type: string
-          description: Description of the assistant. Can be null.
-          example: null
-        avatar:
-          type: string
-          description: URL of the assistant's avatar. Jan-specific property.
-          example: https://pic.png
-        models:
-          type: array
-          description: List of models associated with the assistant. Jan-specific property.
-          items:
-            type: object
-            properties:
-              model_id:
-                type: string
-                example: model_0
-        instructions:
-          type: string
-          description: A system prompt for the assistant.
-          example: Be concise
-        events:
-          type: object
-          description: Event subscription settings for the assistant.
-          properties:
-            in:
-              type: array
-              items:
-                type: string
-            out:
-              type: array
-              items:
-                type: string
-        metadata:
-          type: object
-          description: Metadata associated with the assistant.
-    ListAssistantsResponse: null
-    CreateAssistantResponse:
-      type: object
-      properties:
-        id:
-          type: string
-          description: The identifier of the assistant.
-          example: asst_abc123
-        object:
-          type: string
-          description: Type of the object, indicating it's an assistant.
-          default: assistant
-        version:
-          type: integer
-          description: Version number of the assistant.
-          example: 1
-        created_at:
-          type: integer
-          format: int64
-          description: Unix timestamp representing the creation time of the assistant.
-          example: 1698984975
-        name:
-          type: string
-          description: Name of the assistant.
-          example: Math Tutor
-        description:
-          type: string
-          description: Description of the assistant. Can be null.
-          example: null
-        avatar:
-          type: string
-          description: URL of the assistant's avatar. Jan-specific property.
-          example: https://pic.png
-        models:
-          type: array
-          description: List of models associated with the assistant. Jan-specific property.
-          items:
-            type: object
-            properties:
-              model_id:
-                type: string
-                example: model_0
-        instructions:
-          type: string
-          description: A system prompt for the assistant.
-          example: Be concise
-        events:
-          type: object
-          description: Event subscription settings for the assistant.
-          properties:
-            in:
-              type: array
-              items:
-                type: string
-            out:
-              type: array
-              items:
-                type: string
-        metadata:
-          type: object
-          description: Metadata associated with the assistant.
-    RetrieveAssistantResponse:
-      type: object
-      properties:
-        id:
-          type: string
-          description: The identifier of the assistant.
-          example: asst_abc123
-        object:
-          type: string
-          description: Type of the object, indicating it's an assistant.
-          default: assistant
-        version:
-          type: integer
-          description: Version number of the assistant.
-          example: 1
-        created_at:
-          type: integer
-          format: int64
-          description: Unix timestamp representing the creation time of the assistant.
-          example: 1698984975
-        name:
-          type: string
-          description: Name of the assistant.
-          example: Math Tutor
-        description:
-          type: string
-          description: Description of the assistant. Can be null.
-          example: null
-        avatar:
-          type: string
-          description: URL of the assistant's avatar. Jan-specific property.
-          example: https://pic.png
-        models:
-          type: array
-          description: List of models associated with the assistant. Jan-specific property.
-          items:
-            type: object
-            properties:
-              model_id:
-                type: string
-                example: model_0
-        instructions:
-          type: string
-          description: A system prompt for the assistant.
-          example: Be concise
-        events:
-          type: object
-          description: Event subscription settings for the assistant.
-          properties:
-            in:
-              type: array
-              items:
-                type: string
-            out:
-              type: array
-              items:
-                type: string
-        metadata:
-          type: object
-          description: Metadata associated with the assistant.
-    ModifyAssistantObject:
-      type: object
-      properties:
-        id:
-          type: string
-          description: The identifier of the assistant.
-          example: asst_abc123
-        object:
-          type: string
-          description: Type of the object, indicating it's an assistant.
-          default: assistant
-        version:
-          type: integer
-          description: Version number of the assistant.
-          example: 1
-        created_at:
-          type: integer
-          format: int64
-          description: Unix timestamp representing the creation time of the assistant.
-          example: 1698984975
-        name:
-          type: string
-          description: Name of the assistant.
-          example: Math Tutor
-        description:
-          type: string
-          description: Description of the assistant. Can be null.
-          example: null
-        avatar:
-          type: string
-          description: URL of the assistant's avatar. Jan-specific property.
-          example: https://pic.png
-        models:
-          type: array
-          description: List of models associated with the assistant. Jan-specific property.
-          items:
-            type: object
-            properties:
-              model_id:
-                type: string
-                example: model_0
-        instructions:
-          type: string
-          description: A system prompt for the assistant.
-          example: Be concise
-        events:
-          type: object
-          description: Event subscription settings for the assistant.
-          properties:
-            in:
-              type: array
-              items:
-                type: string
-            out:
-              type: array
-              items:
-                type: string
-        metadata:
-          type: object
-          description: Metadata associated with the assistant.
-    ModifyAssistantResponse:
-      type: object
-      properties:
-        id:
-          type: string
-          description: The identifier of the assistant.
-          example: asst_abc123
-        object:
-          type: string
-          description: Type of the object, indicating it's an assistant.
-          default: assistant
-        version:
-          type: integer
-          description: Version number of the assistant.
-          example: 1
-        created_at:
-          type: integer
-          format: int64
-          description: Unix timestamp representing the creation time of the assistant.
-          example: 1698984975
-        name:
-          type: string
-          description: Name of the assistant.
-          example: Physics Tutor
-        description:
-          type: string
-          description: Description of the assistant. Can be null.
-          example: null
-        avatar:
-          type: string
-          description: URL of the assistant's avatar. Jan-specific property.
-          example: https://pic.png
-        models:
-          type: array
-          description: List of models associated with the assistant. Jan-specific property.
-          items:
-            type: object
-            properties:
-              model_id:
-                type: string
-                example: model_0
-        instructions:
-          type: string
-          description: A system prompt for the assistant.
-          example: Be concise!
-        events:
-          type: object
-          description: Event subscription settings for the assistant.
-          properties:
-            in:
-              type: array
-              items:
-                type: string
-            out:
-              type: array
-              items:
-                type: string
-        metadata:
-          type: object
-          description: Metadata associated with the assistant.
-    DeleteAssistantResponse:
-      type: object
-      properties:
-        id:
-          type: string
-          description: The identifier of the deleted assistant.
-          example: asst_abc123
-        object:
-          type: string
-          description: Type of the object, indicating the assistant has been deleted.
-          example: assistant.deleted
-        deleted:
-          type: boolean
-          description: Indicates whether the assistant was successfully deleted.
-          example: true

docs/openapi/specs/chat.yaml

@@ -1,196 +0,0 @@
----
-components:
-  schemas:
-    ChatObject:
-      type: object
-      properties:
-        messages:
-          type: arrays
-          description: |
-            Contains input data or prompts for the model to process.
-          example:
-            - content: 'Hello there :wave:'
-              role: assistant
-            - content: Can you write a long story
-              role: user
-        stream:
-          type: boolean
-          default: true
-          description:
-            Enables continuous output generation, allowing for streaming of
-            model responses.
-        model:
-          type: string
-          example: gpt-3.5-turbo
-          description: Specifies the model being used for inference or processing tasks.
-        max_tokens:
-          type: number
-          default: 2048
-          description:
-            The maximum number of tokens the model will generate in a single
-            response.
-        stop:
-          type: arrays
-          example:
-            - hello
-          description:
-            Defines specific tokens or phrases at which the model will stop
-            generating further output/
-        frequency_penalty:
-          type: number
-          default: 0
-          description:
-            Adjusts the likelihood of the model repeating words or phrases in
-            its output.
-        presence_penalty:
-          type: number
-          default: 0
-          description:
-            Influences the generation of new and varied concepts in the model's
-            output.
-        temperature:
-          type: number
-          default: 0.7
-          min: 0
-          max: 1
-          description: Controls the randomness of the model's output.
-        top_p:
-          type: number
-          default: 0.95
-          min: 0
-          max: 1
-          description: Set probability threshold for more relevant outputs.
-        cache_prompt:
-          type: boolean
-          default: true
-          description: Optimize performance in repeated or similar requests.
-    ChatCompletionRequest:
-      type: object
-      properties:
-        messages:
-          type: arrays
-          description: |
-            Contains input data or prompts for the model to process.
-          example:
-            - content: You are a helpful assistant.
-              role: system
-            - content: Hello!
-              role: user
-        model:
-          type: string
-          example: tinyllama-1.1b
-          description: |
-            Specifies the model being used for inference or processing tasks.
-        stream:
-          type: boolean
-          default: true
-          description: >
-            Enables continuous output generation, allowing for streaming of
-            model responses.
-        max_tokens:
-          type: number
-          default: 2048
-          description: >
-            The maximum number of tokens the model will generate in a single
-            response.
-        stop:
-          type: arrays
-          example:
-            - hello
-          description: >
-            Defines specific tokens or phrases at which the model will stop
-            generating further output.
-        frequency_penalty:
-          type: number
-          default: 0
-          description: >
-            Adjusts the likelihood of the model repeating words or phrases in
-            its output.
-        presence_penalty:
-          type: number
-          default: 0
-          description: >
-            Influences the generation of new and varied concepts in the model's
-            output.
-        temperature:
-          type: number
-          default: 0.7
-          min: 0
-          max: 1
-          description: |
-            Controls the randomness of the model's output.
-        top_p:
-          type: number
-          default: 0.95
-          min: 0
-          max: 1
-          description: |
-            Set probability threshold for more relevant outputs.
-    ChatCompletionResponse:
-      type: object
-      description: Description of the response structure
-      properties:
-        choices:
-          type: array
-          description: Array of choice objects
-          items:
-            type: object
-            properties:
-              finish_reason:
-                type: string
-                nullable: true
-                example: null
-                description: Reason for finishing the response, if applicable
-              index:
-                type: integer
-                example: 0
-                description: Index of the choice
-              message:
-                type: object
-                properties:
-                  content:
-                    type: string
-                    example: Hello user. What can I help you with?
-                    description: Content of the message
-                  role:
-                    type: string
-                    example: assistant
-                    description: Role of the sender
-        created:
-          type: integer
-          example: 1700193928
-          description: Timestamp of when the response was created
-        id:
-          type: string
-          example: ebwd2niJvJB1Q2Whyvkz
-          description: Unique identifier of the response
-        model:
-          type: string
-          nullable: true
-          example: _
-          description: Model used for generating the response
-        object:
-          type: string
-          example: chat.completion
-          description: Type of the response object
-        system_fingerprint:
-          type: string
-          nullable: true
-          example: _
-          description: System fingerprint
-        usage:
-          type: object
-          description: Information about the usage of tokens
-          properties:
-            completion_tokens:
-              type: integer
-              example: 500
-              description: Number of tokens used for completion
-            prompt_tokens:
-              type: integer
-              example: 33
-              description: Number of tokens used in the prompt
-            total_tokens:
-              type: integer
-              example: 533
-              description: Total number of tokens used

docs/openapi/specs/messages.yaml

@@ -1,313 +0,0 @@
----
-components:
-  schemas:
-    MessageObject:
-      type: object
-      properties:
-        id:
-          type: string
-          description: |
-            Sequential or UUID identifier of the message.
-          example: 0
-        object:
-          type: string
-          description: |
-            Type of the object, defaults to 'thread.message'.
-          example: thread.message
-        created_at:
-          type: integer
-          format: int64
-          description: |
-            Unix timestamp representing the creation time of the message.
-        thread_id:
-          type: string
-          description: >
-            Identifier of the thread to which this message belongs. Defaults to
-            parent thread.
-          example: thread_asdf
-        assistant_id:
-          type: string
-          description: >
-            Identifier of the assistant involved in the message. Defaults to
-            parent thread.
-          example: jan
-        role:
-          type: string
-          enum:
-            - user
-            - assistant
-          description: |
-            Role of the sender, either 'user' or 'assistant'.
-        content:
-          type: array
-          items:
-            type: object
-            properties:
-              type:
-                type: string
-                description: |
-                  Type of content, e.g., 'text'.
-              text:
-                type: object
-                properties:
-                  value:
-                    type: string
-                    description: |
-                      Text content of the message.
-                    example: Hi!?
-                  annotations:
-                    type: array
-                    items:
-                      type: string
-                    description: |
-                      Annotations for the text content, if any.
-                    example: []
-        metadata:
-          type: object
-          description: |
-            Metadata associated with the message, defaults to an empty object.
-          example: {}
-    GetMessageResponse:
-      type: object
-      properties:
-        id:
-          type: string
-          description: The identifier of the message.
-          example: msg_abc123
-        object:
-          type: string
-          description: Type of the object, indicating it's a thread message.
-          default: thread.message
-        created_at:
-          type: integer
-          format: int64
-          description: Unix timestamp representing the creation time of the message.
-          example: 1699017614
-        thread_id:
-          type: string
-          description: Identifier of the thread to which this message belongs.
-          example: thread_abc123
-        role:
-          type: string
-          description: Role of the sender, either 'user' or 'assistant'.
-          example: user
-        content:
-          type: array
-          items:
-            type: object
-            properties:
-              type:
-                type: string
-                description: Type of content, e.g., 'text'.
-                example: text
-              text:
-                type: object
-                properties:
-                  value:
-                    type: string
-                    description: Text content of the message.
-                    example: How does AI work? Explain it in simple terms.
-                  annotations:
-                    type: array
-                    items:
-                      type: string
-                    description: Annotations for the text content, if any.
-                    example: []
-        file_ids:
-          type: array
-          items:
-            type: string
-          description: Array of file IDs associated with the message, if any.
-          example: []
-        assistant_id:
-          type: string
-          description: Identifier of the assistant involved in the message, if applicable.
-          example: null
-        run_id:
-          type: string
-          description: Run ID associated with the message, if applicable.
-          example: null
-        metadata:
-          type: object
-          description: Metadata associated with the message.
-          example: {}
-    CreateMessageResponse:
-      type: object
-      properties:
-        id:
-          type: string
-          description: The identifier of the created message.
-          example: msg_abc123
-        object:
-          type: string
-          description: Type of the object, indicating it's a thread message.
-          example: thread.message
-        created_at:
-          type: integer
-          format: int64
-          description: Unix timestamp representing the creation time of the message.
-          example: 1699017614
-        thread_id:
-          type: string
-          description: Identifier of the thread to which this message belongs.
-          example: thread_abc123
-        role:
-          type: string
-          description: Role of the sender, either 'user' or 'assistant'.
-          example: user
-        content:
-          type: array
-          items:
-            type: object
-            properties:
-              type:
-                type: string
-                description: Type of content, e.g., 'text'.
-                example: text
-              text:
-                type: object
-                properties:
-                  value:
-                    type: string
-                    description: Text content of the message.
-                    example: How does AI work? Explain it in simple terms.
-                  annotations:
-                    type: array
-                    items:
-                      type: string
-                    description: Annotations for the text content, if any.
-                    example: []
-        file_ids:
-          type: array
-          items:
-            type: string
-          description: Array of file IDs associated with the message, if any.
-          example: []
-        assistant_id:
-          type: string
-          description: Identifier of the assistant involved in the message, if applicable.
-          example: null
-        run_id:
-          type: string
-          description: Run ID associated with the message, if applicable.
-          example: null
-        metadata:
-          type: object
-          description: Metadata associated with the message.
-          example: {}
-    ListMessagesResponse:
-      type: object
-      properties:
-        object:
-          type: string
-          description: Type of the object, indicating it's a list.
-          default: list
-        data:
-          type: array
-          items:
-            $ref: '#/components/schemas/ListMessageObject'
-        first_id:
-          type: string
-          description: Identifier of the first message in the list.
-          example: msg_abc123
-        last_id:
-          type: string
-          description: Identifier of the last message in the list.
-          example: msg_abc456
-        has_more:
-          type: boolean
-          description: Indicates whether there are more messages to retrieve.
-          example: false
-    ListMessageObject:
-      type: object
-      properties:
-        id:
-          type: string
-          description: The identifier of the message.
-          example: msg_abc123
-        object:
-          type: string
-          description: Type of the object, indicating it's a thread message.
-          example: thread.message
-        created_at:
-          type: integer
-          format: int64
-          description: Unix timestamp representing the creation time of the message.
-          example: 1699017614
-        thread_id:
-          type: string
-          description: Identifier of the thread to which this message belongs.
-          example: thread_abc123
-        role:
-          type: string
-          description: Role of the sender, either 'user' or 'assistant'.
-          example: user
-        content:
-          type: array
-          items:
-            type: object
-            properties:
-              type:
-                type: string
-                description: Type of content, e.g., 'text'.
-              text:
-                type: object
-                properties:
-                  value:
-                    type: string
-                    description: Text content of the message.
-                    example: How does AI work? Explain it in simple terms.
-                  annotations:
-                    type: array
-                    items:
-                      type: string
-                    description: Annotations for the text content, if any.
-        file_ids:
-          type: array
-          items:
-            type: string
-          description: Array of file IDs associated with the message, if any.
-          example: []
-        assistant_id:
-          type: string
-          description: Identifier of the assistant involved in the message, if applicable.
-          example: null
-        run_id:
-          type: string
-          description: Run ID associated with the message, if applicable.
-          example: null
-        metadata:
-          type: object
-          description: Metadata associated with the message.
-          example: {}
-    MessageFileObject:
-      type: object
-      properties:
-        id:
-          type: string
-          description: The identifier of the file.
-          example: file-abc123
-        object:
-          type: string
-          description: Type of the object, indicating it's a thread message file.
-          example: thread.message.file
-        created_at:
-          type: integer
-          format: int64
-          description: Unix timestamp representing the creation time of the file.
-          example: 1699061776
-        message_id:
-          type: string
-          description: Identifier of the message to which this file is associated.
-          example: msg_abc123
-    ListMessageFilesResponse:
-      type: object
-      properties:
-        object:
-          type: string
-          description: Type of the object, indicating it's a list.
-          default: list
-        data:
-          type: array
-          items:
-            $ref: '#/components/schemas/MessageFileObject'

docs/openapi/specs/models.yaml

@@ -1,259 +0,0 @@
----
-components:
-  schemas:
-    ListModelsResponse:
-      type: object
-      properties:
-        object:
-          type: string
-          enum:
-            - list
-        data:
-          type: array
-          items:
-            $ref: '#/components/schemas/Model'
-      required:
-        - object
-        - data
-    Model:
-      type: object
-      properties:
-        source_url:
-          type: string
-          format: uri
-          description: URL to the source of the model.
-          example: https://huggingface.co/janhq/trinity-v1.2-GGUF/resolve/main/trinity-v1.2.Q4_K_M.gguf
-        id:
-          type: string
-          description:
-            Unique identifier used in chat-completions model_name, matches
-            folder name.
-          example: trinity-v1.2-7b
-        object:
-          type: string
-          example: model
-        name:
-          type: string
-          description: Name of the model.
-          example: Trinity-v1.2 7B Q4
-        version:
-          type: string
-          default: '1.0'
-          description: The version number of the model.
-        description:
-          type: string
-          description: Description of the model.
-          example:
-            Trinity is an experimental model merge using the Slerp method.
-            Recommended for daily assistance purposes.
-        format:
-          type: string
-          description: State format of the model, distinct from the engine.
-          example: gguf
-        settings:
-          type: object
-          properties:
-            ctx_len:
-              type: integer
-              description: Context length.
-              example: 4096
-            prompt_template:
-              type: string
-              example: "<|im_start|>system\n{system_message}<|im_end|>\n<|im_start|>user\n{prompt}<|im_end|>\n<|im_start|>assistant"
-          additionalProperties: false
-        parameters:
-          type: object
-          properties:
-            temperature:
-              example: 0.7
-            top_p:
-              example: 0.95
-            stream:
-              example: true
-            max_tokens:
-              example: 4096
-            stop:
-              example: []
-            frequency_penalty:
-              example: 0
-            presence_penalty:
-              example: 0
-          additionalProperties: false
-        metadata:
-          author:
-            type: string
-            example: Jan
-          tags:
-            example:
-              - 7B
-              - Merged
-              - Featured
-          size:
-            example: 4370000000,
-          cover:
-            example: https://raw.githubusercontent.com/janhq/jan/main/models/trinity-v1.2-7b/cover.png
-        engine:
-          example: nitro
-    ModelObject:
-      type: object
-      properties:
-        id:
-          type: string
-          description: |
-            The identifier of the model.
-          example: trinity-v1.2-7b
-        object:
-          type: string
-          description: |
-            The type of the object, indicating it's a model.
-          default: model
-        created:
-          type: integer
-          format: int64
-          description: |
-            Unix timestamp representing the creation time of the model.
-          example: 1253935178
-        owned_by:
-          type: string
-          description: |
-            The entity that owns the model.
-          example: _
-    GetModelResponse:
-      type: object
-      properties:
-        source_url:
-          type: string
-          format: uri
-          description: URL to the source of the model.
-          example: https://huggingface.co/TheBloke/Mistral-7B-Instruct-v0.2-GGUF/resolve/main/mistral-7b-instruct-v0.2.Q4_K_M.gguf
-        id:
-          type: string
-          description:
-            Unique identifier used in chat-completions model_name, matches
-            folder name.
-          example: mistral-ins-7b-q4
-        object:
-          type: string
-          example: model
-        name:
-          type: string
-          description: Name of the model.
-          example: Mistral Instruct 7B Q4
-        version:
-          type: string
-          default: '1.0'
-          description: The version number of the model.
-        description:
-          type: string
-          description: Description of the model.
-          example:
-            Trinity is an experimental model merge using the Slerp method.
-            Recommended for daily assistance purposes.
-        format:
-          type: string
-          description: State format of the model, distinct from the engine.
-          example: gguf
-        settings:
-          type: object
-          properties:
-            ctx_len:
-              type: integer
-              description: Context length.
-              example: 4096
-            prompt_template:
-              type: string
-              example: '[INST] {prompt} [/INST]'
-          additionalProperties: false
-        parameters:
-          type: object
-          properties:
-            temperature:
-              example: 0.7
-            top_p:
-              example: 0.95
-            stream:
-              example: true
-            max_tokens:
-              example: 4096
-            stop:
-              example: []
-            frequency_penalty:
-              example: 0
-            presence_penalty:
-              example: 0
-          additionalProperties: false
-        metadata:
-          author:
-            type: string
-            example: MistralAI
-          tags:
-            example:
-              - 7B
-              - Featured
-              - Foundation Model
-          size:
-            example: 4370000000,
-          cover:
-            example: https://raw.githubusercontent.com/janhq/jan/main/models/mistral-ins-7b-q4/cover.png
-        engine:
-          example: nitro
-    DeleteModelResponse:
-      type: object
-      properties:
-        id:
-          type: string
-          description: The identifier of the model that was deleted.
-          example: mistral-ins-7b-q4
-        object:
-          type: string
-          description: Type of the object, indicating it's a model.
-          default: model
-        deleted:
-          type: boolean
-          description: Indicates whether the model was successfully deleted.
-          example: true
-    StartModelResponse:
-      type: object
-      properties:
-        id:
-          type: string
-          description: The identifier of the model that was started.
-          example: model-zephyr-7B
-        object:
-          type: string
-          description: Type of the object, indicating it's a model.
-          default: model
-        state:
-          type: string
-          description: The current state of the model after the start operation.
-          example: running
-      required:
-        - id
-        - object
-        - state
-    StopModelResponse:
-      type: object
-      properties:
-        id:
-          type: string
-          description: The identifier of the model that was started.
-          example: model-zephyr-7B
-        object:
-          type: string
-          description: Type of the object, indicating it's a model.
-          default: model
-        state:
-          type: string
-          description: The current state of the model after the start operation.
-          example: stopped
-      required:
-        - id
-        - object
-        - state
-    DownloadModelResponse:
-      type: object
-      properties:
-        message:
-          type: string
-          description: Message indicates Jan starting download corresponding model.
-          example: Starting download mistral-ins-7b-q4

docs/openapi/specs/threads.yaml

@@ -1,227 +0,0 @@
----
-components:
-  schemas:
-    ThreadObject:
-      type: object
-      properties:
-        id:
-          type: string
-          description: |
-            The identifier of the thread, defaults to foldername.
-          example: thread_....
-        object:
-          type: string
-          description: |
-            Type of the object, defaults to thread.
-          example: thread
-        title:
-          type: string
-          description: >
-            A brief summary or description of the thread, defaults to an empty
-            string.
-          example: funny physics joke
-        assistants:
-          type: array
-          description: ''
-          items:
-            properties:
-              assistant_id:
-                type: string
-                description: |
-                  The identifier of assistant, defaults to "jan"
-                example: jan
-              model:
-                type: object
-                properties:
-                  id:
-                    type: string
-                    description: ''
-                    example: ...
-                  settings:
-                    type: object
-                    description: >
-                      Defaults to and overrides assistant.json's "settings" (and if none,
-                      then model.json "settings")
-                  parameters:
-                    type: object
-                    description: >
-                      Defaults to and overrides assistant.json's "parameters" (and if
-                      none, then model.json "parameters")
-        created:
-          type: integer
-          format: int64
-          description: >
-            Unix timestamp representing the creation time of the thread,
-            defaults to file creation time.
-          example: 1231231
-        metadata:
-          type: object
-          description: |
-            Metadata associated with the thread, defaults to an empty object.
-          example: {}
-    GetThreadResponse:
-      type: object
-      properties:
-        id:
-          type: string
-          description: The identifier of the thread.
-          example: thread_abc123
-        object:
-          type: string
-          description: Type of the object
-          example: thread
-        created_at:
-          type: integer
-          format: int64
-          description: Unix timestamp representing the creation time of the thread.
-          example: 1699014083
-        assistants:
-          type: array
-          items:
-            type: string
-          description: List of assistants involved in the thread.
-          example:
-            - assistant-001
-        metadata:
-          type: object
-          description: Metadata associated with the thread.
-          example: {}
-        messages:
-          type: array
-          items:
-            type: string
-          description: List of messages within the thread.
-          example: []
-    CreateThreadResponse:
-      type: object
-      properties:
-        id:
-          type: string
-          description: The identifier of the newly created thread.
-          example: thread_abc123
-        object:
-          type: string
-          description: Type of the object, indicating it's a thread.
-          example: thread
-        created_at:
-          type: integer
-          format: int64
-          description: Unix timestamp representing the creation time of the thread.
-          example: 1699014083
-        metadata:
-          type: object
-          description: Metadata associated with the newly created thread.
-          example: {}
-    CreateThreadObject:
-      type: object
-      properties:
-        object:
-          type: string
-          description: Type of the object, indicating it's a thread.
-          example: thread
-        title:
-          type: string
-          description: >
-            A brief summary or description of the thread, defaults to an empty
-            string.
-          example: funny physics joke
-        assistants:
-          type: array
-          description: assistant involved in the thread
-          items:
-            properties:
-              assistant_id:
-                type: string
-                description: |
-                  The identifier of assistant, defaults to "jan"
-                example: jan
-              assistant_name:
-                type: string
-                description: |
-                  The name of assistant, defaults to "Jan"
-                example: Jan
-              instructions:
-                type: string
-                description: >
-                  The instruction of assistant, defaults to "Be my grammar corrector"
-              model:
-                type: object
-                properties:
-                  id:
-                    type: string
-                    description: Model id
-                    example: mistral-ins-7b-q4
-                  settings:
-                    type: object
-                    description: >
-                      Defaults to and overrides assistant.json's "settings" (and if none,
-                      then model.json "settings")
-                  parameters:
-                    type: object
-                    description: >
-                      Defaults to and overrides assistant.json's "parameters" (and if
-                      none, then model.json "parameters")
-                  engine:
-                    type: string
-                    description: Engine id
-                    example: nitro
-        metadata:
-          type: object
-          description: |
-            Metadata associated with the thread, defaults to an empty object.
-    ThreadMessageObject:
-      type: object
-      properties:
-        role:
-          type: string
-          description: |
-            "Role of the sender, either 'user' or 'assistant'."
-          enum:
-            - user
-            - assistant
-        content:
-          type: string
-          description: |
-            "Text content of the message."
-        file_ids:
-          type: array
-          items:
-            type: string
-          description: |
-            "Array of file IDs associated with the message, if any."
-    ModifyThreadResponse:
-      type: object
-      properties:
-        id:
-          type: string
-          description: |
-            "The identifier of the modified thread."
-          example: thread_abc123
-        object:
-          type: string
-          description: Type of the object, indicating it's a thread.
-          example: thread
-        created_at:
-          type: integer
-          format: int64
-          description: Unix timestamp representing the creation time of the thread.
-          example: 1699014083
-        metadata:
-          type: object
-          description: Metadata associated with the modified thread.
-          example: {}
-    DeleteThreadResponse:
-      type: object
-      properties:
-        id:
-          type: string
-          description: The identifier of the deleted thread.
-          example: thread_abc123
-        object:
-          type: string
-          description: Type of the object, indicating the thread has been deleted.
-          example: thread.deleted
-        deleted:
-          type: boolean
-          description: Indicates whether the thread was successfully deleted.
-          example: true

docs/openapi/version.txt

@@ -1 +0,0 @@
-v1.23.2

docs/src/components/Download/CardDownload.tsx

@@ -18,17 +18,12 @@ type SystemType = {
 
 const systemsTemplate: SystemType[] = [
   {
-    name: 'Mac M1, M2, M3',
-    label: 'Apple Silicon',
+    name: 'Mac ',
+    label: 'Universal',
     logo: FaApple,
-    fileFormat: '{appname}-mac-arm64-{tag}.dmg',
-  },
-  {
-    name: 'Mac (Intel)',
-    label: 'Apple Intel',
-    logo: FaApple,
-    fileFormat: '{appname}-mac-x64-{tag}.dmg',
+    fileFormat: '{appname}-mac-universal-{tag}.dmg',
   },
+
   {
     name: 'Windows',
     label: 'Standard (64-bit)',

docs/src/components/DropdownDownload/index.tsx

@@ -24,14 +24,9 @@ type GpuInfo = {
 
 const systemsTemplate: SystemType[] = [
   {
-    name: 'Download for Mac (M1/M2/M3)',
+    name: 'Download for Mac',
     logo: FaApple,
-    fileFormat: '{appname}-mac-arm64-{tag}.dmg',
-  },
-  {
-    name: 'Download for Mac (Intel)',
-    logo: FaApple,
-    fileFormat: '{appname}-mac-x64-{tag}.dmg',
+    fileFormat: '{appname}-mac-universal-{tag}.dmg',
   },
   {
     name: 'Download for Windows',
@@ -66,27 +61,20 @@ const DropdownDownload = ({ lastRelease }: Props) => {
     type: '',
   })
 
-  const changeDefaultSystem = useCallback(
-    async (systems: SystemType[]) => {
-      const userAgent = navigator.userAgent
-      if (userAgent.includes('Windows')) {
-        // windows user
-        setDefaultSystem(systems[2])
-      } else if (userAgent.includes('Linux')) {
-        // linux user
-        setDefaultSystem(systems[3])
-      } else if (userAgent.includes('Mac OS')) {
-        if (gpuInfo.type === 'Apple Silicon') {
-          setDefaultSystem(systems[0])
-        } else {
-          setDefaultSystem(systems[1])
-        }
-      } else {
-        setDefaultSystem(systems[1])
-      }
-    },
-    [gpuInfo.type]
-  )
+  const changeDefaultSystem = useCallback(async (systems: SystemType[]) => {
+    const userAgent = navigator.userAgent
+    if (userAgent.includes('Windows')) {
+      // windows user
+      setDefaultSystem(systems[2])
+    } else if (userAgent.includes('Linux')) {
+      // linux user
+      setDefaultSystem(systems[3])
+    } else if (userAgent.includes('Mac OS')) {
+      setDefaultSystem(systems[0])
+    } else {
+      setDefaultSystem(systems[1])
+    }
+  }, [])
 
   function getUnmaskedInfo(gl: WebGLRenderingContext): {
     renderer: string

electron/package.json

@@ -15,7 +15,6 @@
       "build/**/*.{js,map}",
       "pre-install",
       "themes",
-      "docs/**/*",
       "scripts/**/*",
       "icons/**/*",
       "themes",

electron/tests/config/fixtures.ts

@@ -108,7 +108,7 @@ export const test = base.extend<
 })
 
 test.beforeAll(async () => {
-  await rmSync(path.join(__dirname, '../../test-data'), {
+  rmSync(path.join(__dirname, '../../test-data'), {
     recursive: true,
     force: true,
   })
@@ -122,6 +122,5 @@ test.beforeAll(async () => {
 })
 
 test.afterAll(async () => {
-  // temporally disabling this due to the config for parallel testing WIP
   // teardownElectron()
 })

electron/tests/e2e/navigation.e2e.spec.ts

@@ -2,11 +2,8 @@ import { expect } from '@playwright/test'
 import { page, test, TIMEOUT } from '../config/fixtures'
 
 test('renders left navigation panel', async () => {
-  const settingsBtn = await page
-    .getByTestId('Thread')
-    .first()
-    .isEnabled({ timeout: TIMEOUT })
-  expect([settingsBtn].filter((e) => !e).length).toBe(0)
+  const threadBtn = page.getByTestId('Thread').first()
+  await expect(threadBtn).toBeVisible({ timeout: TIMEOUT })
   // Chat section should be there
   await page.getByTestId('Local API Server').first().click({
     timeout: TIMEOUT,

electron/tests/e2e/thread.e2e.spec.ts

@@ -15,7 +15,13 @@ test('Select GPT model from Hub and Chat with Invalid API Key', async ({
 
   await page.getByTestId('txt-input-chat').fill('dummy value')
 
-  await page.getByTestId('btn-send-chat').click()
+  const denyButton = page.locator('[data-testid="btn-deny-product-analytics"]')
+
+  if ((await denyButton.count()) > 0) {
+    await denyButton.click({ force: true })
+  } else {
+    await page.getByTestId('btn-send-chat').click({ force: true })
+  }
 
   await page.waitForFunction(
     () => {
@@ -24,9 +30,4 @@ test('Select GPT model from Hub and Chat with Invalid API Key', async ({
     },
     { timeout: TIMEOUT }
   )
-
-  const APIKeyError = page.getByTestId('passthrough-error-message')
-  await expect(APIKeyError).toBeVisible({
-    timeout: TIMEOUT,
-  })
 })

extensions/assistant-extension/src/index.ts

@@ -127,7 +127,7 @@ export default class JanAssistantExtension extends AssistantExtension {
     thread_location: undefined,
     id: 'jan',
     object: 'assistant',
-    created_at: Date.now(),
+    created_at: Date.now() / 1000,
     name: 'Jan',
     description: 'A default assistant that can use all downloaded models',
     model: '*',
@@ -141,7 +141,7 @@ export default class JanAssistantExtension extends AssistantExtension {
           top_k: 2,
           chunk_size: 1024,
           chunk_overlap: 64,
-          retrieval_template: `Use the following pieces of context to answer the question at the end. If you don't know the answer, just say that you don't know, don't try to make up an answer.
+          retrieval_template: `Use the following pieces of context to answer the question at the end.
 ----------------
 CONTEXT: {CONTEXT}
 ----------------

extensions/assistant-extension/src/node/index.ts

@@ -9,13 +9,14 @@ export function toolRetrievalUpdateTextSplitter(
   retrieval.updateTextSplitter(chunkSize, chunkOverlap)
 }
 export async function toolRetrievalIngestNewDocument(
+  thread: string,
   file: string,
   model: string,
   engine: string,
   useTimeWeighted: boolean
 ) {
-  const filePath = path.join(getJanDataFolderPath(), normalizeFilePath(file))
-  const threadPath = path.dirname(filePath.replace('files', ''))
+  const threadPath = path.join(getJanDataFolderPath(), 'threads', thread)
+  const filePath = path.join(getJanDataFolderPath(), 'files', file)
   retrieval.updateEmbeddingEngine(model, engine)
   return retrieval
     .ingestAgentKnowledge(filePath, `${threadPath}/memory`, useTimeWeighted)

extensions/assistant-extension/src/tools/retrieval.ts

@@ -35,6 +35,7 @@ export class RetrievalTool extends InferenceTool {
         await executeOnMain(
           NODE,
           'toolRetrievalIngestNewDocument',
+          data.thread?.id,
           docFile,
           data.model?.id,
           data.model?.engine,

extensions/conversational-extension/package.json

@@ -18,12 +18,14 @@
   "devDependencies": {
     "cpx": "^1.5.0",
     "rimraf": "^3.0.2",
+    "ts-loader": "^9.5.0",
     "webpack": "^5.88.2",
-    "webpack-cli": "^5.1.4",
-    "ts-loader": "^9.5.0"
+    "webpack-cli": "^5.1.4"
   },
   "dependencies": {
-    "@janhq/core": "file:../../core"
+    "@janhq/core": "file:../../core",
+    "ky": "^1.7.2",
+    "p-queue": "^8.0.1"
   },
   "engines": {
     "node": ">=18.0.0"

extensions/conversational-extension/src/@types/global.d.ts

@@ -0,0 +1,14 @@
+export {}
+declare global {
+  declare const API_URL: string
+  declare const SOCKET_URL: string
+
+  interface Core {
+    api: APIFunctions
+    events: EventEmitter
+  }
+  interface Window {
+    core?: Core | undefined
+    electronAPI?: any | undefined
+  }
+}

extensions/conversational-extension/src/Conversational.test.ts

@@ -1,408 +0,0 @@
-/**
- * @jest-environment jsdom
- */
-jest.mock('@janhq/core', () => ({
-  ...jest.requireActual('@janhq/core/node'),
-  fs: {
-    existsSync: jest.fn(),
-    mkdir: jest.fn(),
-    writeFileSync: jest.fn(),
-    readdirSync: jest.fn(),
-    readFileSync: jest.fn(),
-    appendFileSync: jest.fn(),
-    rm: jest.fn(),
-    writeBlob: jest.fn(),
-    joinPath: jest.fn(),
-    fileStat: jest.fn(),
-  },
-  joinPath: jest.fn(),
-  ConversationalExtension: jest.fn(),
-}))
-
-import { fs } from '@janhq/core'
-
-import JSONConversationalExtension from '.'
-
-describe('JSONConversationalExtension Tests', () => {
-  let extension: JSONConversationalExtension
-
-  beforeEach(() => {
-    // @ts-ignore
-    extension = new JSONConversationalExtension()
-  })
-
-  it('should create thread folder on load if it does not exist', async () => {
-    // @ts-ignore
-    jest.spyOn(fs, 'existsSync').mockResolvedValue(false)
-    const mkdirSpy = jest.spyOn(fs, 'mkdir').mockResolvedValue({})
-
-    await extension.onLoad()
-
-    expect(mkdirSpy).toHaveBeenCalledWith('file://threads')
-  })
-
-  it('should log message on unload', () => {
-    const consoleSpy = jest.spyOn(console, 'debug').mockImplementation()
-
-    extension.onUnload()
-
-    expect(consoleSpy).toHaveBeenCalledWith(
-      'JSONConversationalExtension unloaded'
-    )
-  })
-
-  it('should return sorted threads', async () => {
-    jest
-      .spyOn(extension, 'getValidThreadDirs')
-      .mockResolvedValue(['dir1', 'dir2'])
-    jest
-      .spyOn(extension, 'readThread')
-      .mockResolvedValueOnce({ updated: '2023-01-01' })
-      .mockResolvedValueOnce({ updated: '2023-01-02' })
-
-    const threads = await extension.getThreads()
-
-    expect(threads).toEqual([
-      { updated: '2023-01-02' },
-      { updated: '2023-01-01' },
-    ])
-  })
-
-  it('should ignore broken threads', async () => {
-    jest
-      .spyOn(extension, 'getValidThreadDirs')
-      .mockResolvedValue(['dir1', 'dir2'])
-    jest
-      .spyOn(extension, 'readThread')
-      .mockResolvedValueOnce(JSON.stringify({ updated: '2023-01-01' }))
-      .mockResolvedValueOnce('this_is_an_invalid_json_content')
-
-    const threads = await extension.getThreads()
-
-    expect(threads).toEqual([{ updated: '2023-01-01' }])
-  })
-
-  it('should save thread', async () => {
-    // @ts-ignore
-    jest.spyOn(fs, 'existsSync').mockResolvedValue(false)
-    const mkdirSpy = jest.spyOn(fs, 'mkdir').mockResolvedValue({})
-    const writeFileSyncSpy = jest
-      .spyOn(fs, 'writeFileSync')
-      .mockResolvedValue({})
-
-    const thread = { id: '1', updated: '2023-01-01' } as any
-    await extension.saveThread(thread)
-
-    expect(mkdirSpy).toHaveBeenCalled()
-    expect(writeFileSyncSpy).toHaveBeenCalled()
-  })
-
-  it('should delete thread', async () => {
-    const rmSpy = jest.spyOn(fs, 'rm').mockResolvedValue({})
-
-    await extension.deleteThread('1')
-
-    expect(rmSpy).toHaveBeenCalled()
-  })
-
-  it('should add new message', async () => {
-    // @ts-ignore
-    jest.spyOn(fs, 'existsSync').mockResolvedValue(false)
-    const mkdirSpy = jest.spyOn(fs, 'mkdir').mockResolvedValue({})
-    const appendFileSyncSpy = jest
-      .spyOn(fs, 'appendFileSync')
-      .mockResolvedValue({})
-
-    const message = {
-      thread_id: '1',
-      content: [{ type: 'text', text: { annotations: [] } }],
-    } as any
-    await extension.addNewMessage(message)
-
-    expect(mkdirSpy).toHaveBeenCalled()
-    expect(appendFileSyncSpy).toHaveBeenCalled()
-  })
-
-  it('should store image', async () => {
-    const writeBlobSpy = jest.spyOn(fs, 'writeBlob').mockResolvedValue({})
-
-    await extension.storeImage(
-      'data:image/png;base64,abcd',
-      'path/to/image.png'
-    )
-
-    expect(writeBlobSpy).toHaveBeenCalled()
-  })
-
-  it('should store file', async () => {
-    const writeBlobSpy = jest.spyOn(fs, 'writeBlob').mockResolvedValue({})
-
-    await extension.storeFile(
-      'data:application/pdf;base64,abcd',
-      'path/to/file.pdf'
-    )
-
-    expect(writeBlobSpy).toHaveBeenCalled()
-  })
-
-  it('should write messages', async () => {
-    // @ts-ignore
-    jest.spyOn(fs, 'existsSync').mockResolvedValue(false)
-    const mkdirSpy = jest.spyOn(fs, 'mkdir').mockResolvedValue({})
-    const writeFileSyncSpy = jest
-      .spyOn(fs, 'writeFileSync')
-      .mockResolvedValue({})
-
-    const messages = [{ id: '1', thread_id: '1', content: [] }] as any
-    await extension.writeMessages('1', messages)
-
-    expect(mkdirSpy).toHaveBeenCalled()
-    expect(writeFileSyncSpy).toHaveBeenCalled()
-  })
-
-  it('should get all messages on string response', async () => {
-    jest.spyOn(fs, 'readdirSync').mockResolvedValue(['messages.jsonl'])
-    jest.spyOn(fs, 'readFileSync').mockResolvedValue('{"id":"1"}\n{"id":"2"}\n')
-
-    const messages = await extension.getAllMessages('1')
-
-    expect(messages).toEqual([{ id: '1' }, { id: '2' }])
-  })
-
-  it('should get all messages on object response', async () => {
-    jest.spyOn(fs, 'readdirSync').mockResolvedValue(['messages.jsonl'])
-    jest.spyOn(fs, 'readFileSync').mockResolvedValue({ id: 1 })
-
-    const messages = await extension.getAllMessages('1')
-
-    expect(messages).toEqual([{ id: 1 }])
-  })
-
-  it('get all messages return empty on error', async () => {
-    jest.spyOn(fs, 'readdirSync').mockRejectedValue(['messages.jsonl'])
-
-    const messages = await extension.getAllMessages('1')
-
-    expect(messages).toEqual([])
-  })
-
-  it('return empty messages on no messages file', async () => {
-    jest.spyOn(fs, 'readdirSync').mockResolvedValue([])
-
-    const messages = await extension.getAllMessages('1')
-
-    expect(messages).toEqual([])
-  })
-
-  it('should ignore error message', async () => {
-    jest.spyOn(fs, 'readdirSync').mockResolvedValue(['messages.jsonl'])
-    jest
-      .spyOn(fs, 'readFileSync')
-      .mockResolvedValue('{"id":"1"}\nyolo\n{"id":"2"}\n')
-
-    const messages = await extension.getAllMessages('1')
-
-    expect(messages).toEqual([{ id: '1' }, { id: '2' }])
-  })
-
-  it('should create thread folder on load if it does not exist', async () => {
-    // @ts-ignore
-    jest.spyOn(fs, 'existsSync').mockResolvedValue(false)
-    const mkdirSpy = jest.spyOn(fs, 'mkdir').mockResolvedValue({})
-
-    await extension.onLoad()
-
-    expect(mkdirSpy).toHaveBeenCalledWith('file://threads')
-  })
-
-  it('should log message on unload', () => {
-    const consoleSpy = jest.spyOn(console, 'debug').mockImplementation()
-
-    extension.onUnload()
-
-    expect(consoleSpy).toHaveBeenCalledWith(
-      'JSONConversationalExtension unloaded'
-    )
-  })
-
-  it('should return sorted threads', async () => {
-    jest
-      .spyOn(extension, 'getValidThreadDirs')
-      .mockResolvedValue(['dir1', 'dir2'])
-    jest
-      .spyOn(extension, 'readThread')
-      .mockResolvedValueOnce({ updated: '2023-01-01' })
-      .mockResolvedValueOnce({ updated: '2023-01-02' })
-
-    const threads = await extension.getThreads()
-
-    expect(threads).toEqual([
-      { updated: '2023-01-02' },
-      { updated: '2023-01-01' },
-    ])
-  })
-
-  it('should ignore broken threads', async () => {
-    jest
-      .spyOn(extension, 'getValidThreadDirs')
-      .mockResolvedValue(['dir1', 'dir2'])
-    jest
-      .spyOn(extension, 'readThread')
-      .mockResolvedValueOnce(JSON.stringify({ updated: '2023-01-01' }))
-      .mockResolvedValueOnce('this_is_an_invalid_json_content')
-
-    const threads = await extension.getThreads()
-
-    expect(threads).toEqual([{ updated: '2023-01-01' }])
-  })
-
-  it('should save thread', async () => {
-    // @ts-ignore
-    jest.spyOn(fs, 'existsSync').mockResolvedValue(false)
-    const mkdirSpy = jest.spyOn(fs, 'mkdir').mockResolvedValue({})
-    const writeFileSyncSpy = jest
-      .spyOn(fs, 'writeFileSync')
-      .mockResolvedValue({})
-
-    const thread = { id: '1', updated: '2023-01-01' } as any
-    await extension.saveThread(thread)
-
-    expect(mkdirSpy).toHaveBeenCalled()
-    expect(writeFileSyncSpy).toHaveBeenCalled()
-  })
-
-  it('should delete thread', async () => {
-    const rmSpy = jest.spyOn(fs, 'rm').mockResolvedValue({})
-
-    await extension.deleteThread('1')
-
-    expect(rmSpy).toHaveBeenCalled()
-  })
-
-  it('should add new message', async () => {
-    // @ts-ignore
-    jest.spyOn(fs, 'existsSync').mockResolvedValue(false)
-    const mkdirSpy = jest.spyOn(fs, 'mkdir').mockResolvedValue({})
-    const appendFileSyncSpy = jest
-      .spyOn(fs, 'appendFileSync')
-      .mockResolvedValue({})
-
-    const message = {
-      thread_id: '1',
-      content: [{ type: 'text', text: { annotations: [] } }],
-    } as any
-    await extension.addNewMessage(message)
-
-    expect(mkdirSpy).toHaveBeenCalled()
-    expect(appendFileSyncSpy).toHaveBeenCalled()
-  })
-
-  it('should add new image message', async () => {
-    jest
-      .spyOn(fs, 'existsSync')
-      // @ts-ignore
-      .mockResolvedValueOnce(false)
-      // @ts-ignore
-      .mockResolvedValueOnce(false)
-      // @ts-ignore
-      .mockResolvedValueOnce(true)
-    const mkdirSpy = jest.spyOn(fs, 'mkdir').mockResolvedValue({})
-    const appendFileSyncSpy = jest
-      .spyOn(fs, 'appendFileSync')
-      .mockResolvedValue({})
-    jest.spyOn(fs, 'writeBlob').mockResolvedValue({})
-
-    const message = {
-      thread_id: '1',
-      content: [
-        { type: 'image', text: { annotations: ['data:image;base64,hehe'] } },
-      ],
-    } as any
-    await extension.addNewMessage(message)
-
-    expect(mkdirSpy).toHaveBeenCalled()
-    expect(appendFileSyncSpy).toHaveBeenCalled()
-  })
-
-  it('should add new pdf message', async () => {
-    jest
-      .spyOn(fs, 'existsSync')
-      // @ts-ignore
-      .mockResolvedValueOnce(false)
-      // @ts-ignore
-      .mockResolvedValueOnce(false)
-      // @ts-ignore
-      .mockResolvedValueOnce(true)
-    const mkdirSpy = jest.spyOn(fs, 'mkdir').mockResolvedValue({})
-    const appendFileSyncSpy = jest
-      .spyOn(fs, 'appendFileSync')
-      .mockResolvedValue({})
-    jest.spyOn(fs, 'writeBlob').mockResolvedValue({})
-
-    const message = {
-      thread_id: '1',
-      content: [
-        { type: 'pdf', text: { annotations: ['data:pdf;base64,hehe'] } },
-      ],
-    } as any
-    await extension.addNewMessage(message)
-
-    expect(mkdirSpy).toHaveBeenCalled()
-    expect(appendFileSyncSpy).toHaveBeenCalled()
-  })
-
-  it('should store image', async () => {
-    const writeBlobSpy = jest.spyOn(fs, 'writeBlob').mockResolvedValue({})
-
-    await extension.storeImage(
-      'data:image/png;base64,abcd',
-      'path/to/image.png'
-    )
-
-    expect(writeBlobSpy).toHaveBeenCalled()
-  })
-
-  it('should store file', async () => {
-    const writeBlobSpy = jest.spyOn(fs, 'writeBlob').mockResolvedValue({})
-
-    await extension.storeFile(
-      'data:application/pdf;base64,abcd',
-      'path/to/file.pdf'
-    )
-
-    expect(writeBlobSpy).toHaveBeenCalled()
-  })
-})
-
-describe('test readThread', () => {
-  let extension: JSONConversationalExtension
-
-  beforeEach(() => {
-    // @ts-ignore
-    extension = new JSONConversationalExtension()
-  })
-
-  it('should read thread', async () => {
-    jest
-      .spyOn(fs, 'readFileSync')
-      .mockResolvedValue(JSON.stringify({ id: '1' }))
-    const thread = await extension.readThread('1')
-    expect(thread).toEqual(`{"id":"1"}`)
-  })
-
-  it('getValidThreadDirs should return valid thread directories', async () => {
-    jest
-      .spyOn(fs, 'readdirSync')
-      .mockResolvedValueOnce(['1', '2', '3'])
-      .mockResolvedValueOnce(['thread.json'])
-      .mockResolvedValueOnce(['thread.json'])
-      .mockResolvedValueOnce([])
-      // @ts-ignore
-    jest.spyOn(fs, 'existsSync').mockResolvedValue(true)
-    jest.spyOn(fs, 'fileStat').mockResolvedValue({
-      isDirectory: true,
-    } as any)
-    const validThreadDirs = await extension.getValidThreadDirs()
-    expect(validThreadDirs).toEqual(['1', '2'])
-  })
-})

extensions/conversational-extension/src/index.ts

@@ -1,90 +1,71 @@
 import {
-  fs,
-  joinPath,
   ConversationalExtension,
   Thread,
+  ThreadAssistantInfo,
   ThreadMessage,
 } from '@janhq/core'
-import { safelyParseJSON } from './jsonUtil'
+import ky from 'ky'
+import PQueue from 'p-queue'
+
+type ThreadList = {
+  data: Thread[]
+}
+
+type MessageList = {
+  data: ThreadMessage[]
+}
 
 /**
  * JSONConversationalExtension is a ConversationalExtension implementation that provides
  * functionality for managing threads.
  */
-export default class JSONConversationalExtension extends ConversationalExtension {
-  private static readonly _threadFolder = 'file://threads'
-  private static readonly _threadInfoFileName = 'thread.json'
-  private static readonly _threadMessagesFileName = 'messages.jsonl'
+export default class CortexConversationalExtension extends ConversationalExtension {
+  queue = new PQueue({ concurrency: 1 })
 
   /**
    * Called when the extension is loaded.
    */
   async onLoad() {
-    if (!(await fs.existsSync(JSONConversationalExtension._threadFolder))) {
-      await fs.mkdir(JSONConversationalExtension._threadFolder)
-    }
+    this.queue.add(() => this.healthz())
   }
 
   /**
    * Called when the extension is unloaded.
    */
-  onUnload() {
-    console.debug('JSONConversationalExtension unloaded')
-  }
+  onUnload() {}
 
   /**
    * Returns a Promise that resolves to an array of Conversation objects.
    */
-  async getThreads(): Promise<Thread[]> {
-    try {
-      const threadDirs = await this.getValidThreadDirs()
-
-      const promises = threadDirs.map((dirName) => this.readThread(dirName))
-      const promiseResults = await Promise.allSettled(promises)
-      const convos = promiseResults
-        .map((result) => {
-          if (result.status === 'fulfilled') {
-            return typeof result.value === 'object'
-              ? result.value
-              : safelyParseJSON(result.value)
-          }
-          return undefined
-        })
-        .filter((convo) => !!convo)
-      convos.sort(
-        (a, b) => new Date(b.updated).getTime() - new Date(a.updated).getTime()
-      )
-
-      return convos
-    } catch (error) {
-      console.error(error)
-      return []
-    }
+  async listThreads(): Promise<Thread[]> {
+    return this.queue.add(() =>
+      ky
+        .get(`${API_URL}/v1/threads`)
+        .json<ThreadList>()
+        .then((e) => e.data)
+    ) as Promise<Thread[]>
   }
 
   /**
    * Saves a Thread object to a json file.
    * @param thread The Thread object to save.
    */
-  async saveThread(thread: Thread): Promise<void> {
-    try {
-      const threadDirPath = await joinPath([
-        JSONConversationalExtension._threadFolder,
-        thread.id,
-      ])
-      const threadJsonPath = await joinPath([
-        threadDirPath,
-        JSONConversationalExtension._threadInfoFileName,
-      ])
-      if (!(await fs.existsSync(threadDirPath))) {
-        await fs.mkdir(threadDirPath)
-      }
+  async createThread(thread: Thread): Promise<Thread> {
+    return this.queue.add(() =>
+      ky.post(`${API_URL}/v1/threads`, { json: thread }).json<Thread>()
+    ) as Promise<Thread>
+  }
 
-      await fs.writeFileSync(threadJsonPath, JSON.stringify(thread, null, 2))
-    } catch (err) {
-      console.error(err)
-      Promise.reject(err)
-    }
+  /**
+   * Saves a Thread object to a json file.
+   * @param thread The Thread object to save.
+   */
+  async modifyThread(thread: Thread): Promise<void> {
+    return this.queue
+      .add(() =>
+        ky.post(`${API_URL}/v1/threads/${thread.id}`, { json: thread })
+      )
+      .then()
   }
 
   /**
@@ -92,189 +73,126 @@ export default class JSONConversationalExtension extends ConversationalExtension
    * @param threadId The ID of the thread to delete.
    */
   async deleteThread(threadId: string): Promise<void> {
-    const path = await joinPath([
-      JSONConversationalExtension._threadFolder,
-      `${threadId}`,
-    ])
-    try {
-      await fs.rm(path)
-    } catch (err) {
-      console.error(err)
-    }
+    return this.queue
+      .add(() => ky.delete(`${API_URL}/v1/threads/${threadId}`))
+      .then()
   }
 
-  async addNewMessage(message: ThreadMessage): Promise<void> {
-    try {
-      const threadDirPath = await joinPath([
-        JSONConversationalExtension._threadFolder,
-        message.thread_id,
-      ])
-      const threadMessagePath = await joinPath([
-        threadDirPath,
-        JSONConversationalExtension._threadMessagesFileName,
-      ])
-      if (!(await fs.existsSync(threadDirPath))) await fs.mkdir(threadDirPath)
-
-      if (message.content[0]?.type === 'image') {
-        const filesPath = await joinPath([threadDirPath, 'files'])
-        if (!(await fs.existsSync(filesPath))) await fs.mkdir(filesPath)
-
-        const imagePath = await joinPath([filesPath, `${message.id}.png`])
-        const base64 = message.content[0].text.annotations[0]
-        await this.storeImage(base64, imagePath)
-        if ((await fs.existsSync(imagePath)) && message.content?.length) {
-          // Use file path instead of blob
-          message.content[0].text.annotations[0] = `threads/${message.thread_id}/files/${message.id}.png`
-        }
-      }
-
-      if (message.content[0]?.type === 'pdf') {
-        const filesPath = await joinPath([threadDirPath, 'files'])
-        if (!(await fs.existsSync(filesPath))) await fs.mkdir(filesPath)
-
-        const filePath = await joinPath([filesPath, `${message.id}.pdf`])
-        const blob = message.content[0].text.annotations[0]
-        await this.storeFile(blob, filePath)
-
-        if ((await fs.existsSync(filePath)) && message.content?.length) {
-          // Use file path instead of blob
-          message.content[0].text.annotations[0] = `threads/${message.thread_id}/files/${message.id}.pdf`
-        }
-      }
-      await fs.appendFileSync(threadMessagePath, JSON.stringify(message) + '\n')
-      Promise.resolve()
-    } catch (err) {
-      Promise.reject(err)
-    }
+  /**
+   * Adds a new message to a specified thread.
+   * @param message The ThreadMessage object to be added.
+   * @returns A Promise that resolves when the message has been added.
+   */
+  async createMessage(message: ThreadMessage): Promise<ThreadMessage> {
+    return this.queue.add(() =>
+      ky
+        .post(`${API_URL}/v1/threads/${message.thread_id}/messages`, {
+          json: message,
+        })
+        .json<ThreadMessage>()
+    ) as Promise<ThreadMessage>
   }
 
-  async storeImage(base64: string, filePath: string): Promise<void> {
-    const base64Data = base64.replace(/^data:image\/\w+;base64,/, '')
-
-    try {
-      await fs.writeBlob(filePath, base64Data)
-    } catch (err) {
-      console.error(err)
-    }
+  /**
+   * Modifies a message in a thread.
+   * @param message
+   * @returns
+   */
+  async modifyMessage(message: ThreadMessage): Promise<ThreadMessage> {
+    return this.queue.add(() =>
+      ky
+        .post(
+          `${API_URL}/v1/threads/${message.thread_id}/messages/${message.id}`,
+          {
+            json: message,
+          }
+        )
+        .json<ThreadMessage>()
+    ) as Promise<ThreadMessage>
   }
 
-  async storeFile(base64: string, filePath: string): Promise<void> {
-    const base64Data = base64.replace(/^data:application\/pdf;base64,/, '')
-    try {
-      await fs.writeBlob(filePath, base64Data)
-    } catch (err) {
-      console.error(err)
-    }
+  /**
+   * Deletes a specific message from a thread.
+   * @param threadId The ID of the thread containing the message.
+   * @param messageId The ID of the message to be deleted.
+   * @returns A Promise that resolves when the message has been successfully deleted.
+   */
+  async deleteMessage(threadId: string, messageId: string): Promise<void> {
+    return this.queue
+      .add(() =>
+        ky.delete(`${API_URL}/v1/threads/${threadId}/messages/${messageId}`)
+      )
+      .then()
   }
 
-  async writeMessages(
+  /**
+   * Retrieves all messages for a specified thread.
+   * @param threadId The ID of the thread to get messages from.
+   * @returns A Promise that resolves to an array of ThreadMessage objects.
+   */
+  async listMessages(threadId: string): Promise<ThreadMessage[]> {
+    return this.queue.add(() =>
+      ky
+        .get(`${API_URL}/v1/threads/${threadId}/messages?order=asc`)
+        .json<MessageList>()
+        .then((e) => e.data)
+    ) as Promise<ThreadMessage[]>
+  }
+
+  /**
+   * Retrieves the assistant information for a specified thread.
+   * @param threadId The ID of the thread for which to retrieve assistant information.
+   * @returns A Promise that resolves to a ThreadAssistantInfo object containing
+   * the details of the assistant associated with the specified thread.
+   */
+  async getThreadAssistant(threadId: string): Promise<ThreadAssistantInfo> {
+    return this.queue.add(() =>
+      ky.get(`${API_URL}/v1/assistants/${threadId}`).json<ThreadAssistantInfo>()
+    ) as Promise<ThreadAssistantInfo>
+  }
+  /**
+   * Creates a new assistant for the specified thread.
+   * @param threadId The ID of the thread for which the assistant is being created.
+   * @param assistant The information about the assistant to be created.
+   * @returns A Promise that resolves to the newly created ThreadAssistantInfo object.
+   */
+  async createThreadAssistant(
     threadId: string,
-    messages: ThreadMessage[]
-  ): Promise<void> {
-    try {
-      const threadDirPath = await joinPath([
-        JSONConversationalExtension._threadFolder,
-        threadId,
-      ])
-      const threadMessagePath = await joinPath([
-        threadDirPath,
-        JSONConversationalExtension._threadMessagesFileName,
-      ])
-      if (!(await fs.existsSync(threadDirPath))) await fs.mkdir(threadDirPath)
-      await fs.writeFileSync(
-        threadMessagePath,
-        messages.map((msg) => JSON.stringify(msg)).join('\n') +
-          (messages.length ? '\n' : '')
-      )
-      Promise.resolve()
-    } catch (err) {
-      Promise.reject(err)
-    }
+    assistant: ThreadAssistantInfo
+  ): Promise<ThreadAssistantInfo> {
+    return this.queue.add(() =>
+      ky
+        .post(`${API_URL}/v1/assistants/${threadId}`, { json: assistant })
+        .json<ThreadAssistantInfo>()
+    ) as Promise<ThreadAssistantInfo>
   }
 
   /**
-   * A promise builder for reading a thread from a file.
-   * @param threadDirName the thread dir we are reading from.
-   * @returns data of the thread
+   * Modifies an existing assistant for the specified thread.
+   * @param threadId The ID of the thread for which the assistant is being modified.
+   * @param assistant The updated information for the assistant.
+   * @returns A Promise that resolves to the updated ThreadAssistantInfo object.
    */
-  async readThread(threadDirName: string): Promise<any> {
-    return fs.readFileSync(
-      await joinPath([
-        JSONConversationalExtension._threadFolder,
-        threadDirName,
-        JSONConversationalExtension._threadInfoFileName,
-      ]),
-      'utf-8'
-    )
+  async modifyThreadAssistant(
+    threadId: string,
+    assistant: ThreadAssistantInfo
+  ): Promise<ThreadAssistantInfo> {
+    return this.queue.add(() =>
+      ky
+        .patch(`${API_URL}/v1/assistants/${threadId}`, { json: assistant })
+        .json<ThreadAssistantInfo>()
+    ) as Promise<ThreadAssistantInfo>
   }
 
   /**
-   * Returns a Promise that resolves to an array of thread directories.
-   * @private
+   * Do health check on cortex.cpp
+   * @returns
    */
-  async getValidThreadDirs(): Promise<string[]> {
-    const fileInsideThread: string[] = await fs.readdirSync(
-      JSONConversationalExtension._threadFolder
-    )
-
-    const threadDirs: string[] = []
-    for (let i = 0; i < fileInsideThread.length; i++) {
-      const path = await joinPath([
-        JSONConversationalExtension._threadFolder,
-        fileInsideThread[i],
-      ])
-      if (!(await fs.fileStat(path))?.isDirectory) continue
-
-      const isHavingThreadInfo = (await fs.readdirSync(path)).includes(
-        JSONConversationalExtension._threadInfoFileName
-      )
-      if (!isHavingThreadInfo) {
-        console.debug(`Ignore ${path} because it does not have thread info`)
-        continue
-      }
-
-      threadDirs.push(fileInsideThread[i])
-    }
-    return threadDirs
-  }
-
-  async getAllMessages(threadId: string): Promise<ThreadMessage[]> {
-    try {
-      const threadDirPath = await joinPath([
-        JSONConversationalExtension._threadFolder,
-        threadId,
-      ])
-
-      const files: string[] = await fs.readdirSync(threadDirPath)
-      if (
-        !files.includes(JSONConversationalExtension._threadMessagesFileName)
-      ) {
-        console.debug(`${threadDirPath} not contains message file`)
-        return []
-      }
-
-      const messageFilePath = await joinPath([
-        threadDirPath,
-        JSONConversationalExtension._threadMessagesFileName,
-      ])
-
-      let readResult = await fs.readFileSync(messageFilePath, 'utf-8')
-
-      if (typeof readResult === 'object') {
-        readResult = JSON.stringify(readResult)
-      }
-
-      const result = readResult.split('\n').filter((line) => line !== '')
-
-      const messages: ThreadMessage[] = []
-      result.forEach((line: string) => {
-        const message = safelyParseJSON(line)
-        if (message) messages.push(safelyParseJSON(line))
+  healthz(): Promise<void> {
+    return ky
+      .get(`${API_URL}/healthz`, {
+        retry: { limit: 20, delay: () => 500, methods: ['get'] },
       })
-      return messages
-    } catch (err) {
-      console.error(err)
-      return []
-    }
+      .then(() => {})
   }
 }

extensions/conversational-extension/src/jsonUtil.ts

@@ -1,14 +0,0 @@
-// Note about performance
-// The v8 JavaScript engine used by Node.js cannot optimise functions which contain a try/catch block.
-// v8 4.5 and above can optimise try/catch
-export function safelyParseJSON(json) {
-  // This function cannot be optimised, it's best to
-  // keep it small!
-  var parsed
-  try {
-    parsed = JSON.parse(json)
-  } catch (e) {
-    return undefined
-  }
-  return parsed // Could be undefined!
-}

extensions/conversational-extension/webpack.config.js

@@ -17,7 +17,12 @@ module.exports = {
     filename: 'index.js', // Adjust the output file name as needed
     library: { type: 'module' }, // Specify ESM output format
   },
-  plugins: [new webpack.DefinePlugin({})],
+  plugins: [
+    new webpack.DefinePlugin({
+      API_URL: JSON.stringify('http://127.0.0.1:39291'),
+      SOCKET_URL: JSON.stringify('ws://127.0.0.1:39291'),
+    }),
+  ],
   resolve: {
     extensions: ['.ts', '.js'],
   },

extensions/inference-cortex-extension/bin/version.txt

@@ -1 +1 @@
-1.0.4
+1.0.5-rc2

extensions/inference-cortex-extension/download.bat

@@ -2,7 +2,7 @@
 set BIN_PATH=./bin
 set SHARED_PATH=./../../electron/shared
 set /p CORTEX_VERSION=<./bin/version.txt
-set ENGINE_VERSION=0.1.40
+set ENGINE_VERSION=0.1.42
 
 @REM Download cortex.llamacpp binaries
 set DOWNLOAD_URL=https://github.com/janhq/cortex.llamacpp/releases/download/v%ENGINE_VERSION%/cortex.llamacpp-%ENGINE_VERSION%-windows-amd64
@@ -38,4 +38,4 @@ for %%F in (%SUBFOLDERS%) do (
     )
 )
 
-echo DLL files moved successfully.
+echo DLL files moved successfully.

extensions/inference-cortex-extension/download.sh

@@ -2,7 +2,7 @@
 
 # Read CORTEX_VERSION
 CORTEX_VERSION=$(cat ./bin/version.txt)
-ENGINE_VERSION=0.1.40
+ENGINE_VERSION=0.1.42
 CORTEX_RELEASE_URL="https://github.com/janhq/cortex.cpp/releases/download"
 ENGINE_DOWNLOAD_URL="https://github.com/janhq/cortex.llamacpp/releases/download/v${ENGINE_VERSION}/cortex.llamacpp-${ENGINE_VERSION}"
 CUDA_DOWNLOAD_URL="https://github.com/janhq/cortex.llamacpp/releases/download/v${ENGINE_VERSION}"

extensions/inference-cortex-extension/rollup.config.ts

@@ -120,7 +120,7 @@ export default [
         SETTINGS: JSON.stringify(defaultSettingJson),
         CORTEX_API_URL: JSON.stringify('http://127.0.0.1:39291'),
         CORTEX_SOCKET_URL: JSON.stringify('ws://127.0.0.1:39291'),
-        CORTEX_ENGINE_VERSION: JSON.stringify('v0.1.40'),
+        CORTEX_ENGINE_VERSION: JSON.stringify('v0.1.42'),
       }),
       // Allow json resolution
       json(),

extensions/inference-groq-extension/resources/models.json

@@ -61,37 +61,6 @@
     },
     "engine": "groq"
   },
-  {
-    "sources": [
-      {
-        "url": "https://groq.com"
-      }
-    ],
-    "id": "llama-3.1-70b-versatile",
-    "object": "model",
-    "name": "Groq Llama 3.1 70b Versatile",
-    "version": "1.1",
-    "description": "Groq Llama 3.1 70b Versatile with supercharged speed!",
-    "format": "api",
-    "settings": {},
-    "parameters": {
-      "max_tokens": 8000,
-      "temperature": 0.7,
-      "top_p": 0.95,
-      "stream": true,
-      "stop": [],
-      "frequency_penalty": 0,
-      "presence_penalty": 0
-    },
-    "metadata": {
-      "author": "Meta",
-      "tags": [
-        "General",
-        "Big Context Length"
-      ]
-    },
-    "engine": "groq"
-  },
   {
     "sources": [
       {

extensions/inference-openai-extension/package.json

@@ -1,7 +1,7 @@
 {
   "name": "@janhq/inference-openai-extension",
   "productName": "OpenAI Inference Engine",
-  "version": "1.0.4",
+  "version": "1.0.5",
   "description": "This extension enables OpenAI chat completion API calls",
   "main": "dist/index.js",
   "module": "dist/module.js",

extensions/inference-openai-extension/resources/models.json

@@ -67,7 +67,9 @@
     "version": "1.1",
     "description": "OpenAI GPT 4o is a new flagship model with fast speed and high quality",
     "format": "api",
-    "settings": {},
+    "settings": {
+      "vision_model": true
+    },
     "parameters": {
       "max_tokens": 4096,
       "temperature": 0.7,
@@ -97,10 +99,10 @@
     "format": "api",
     "settings": {},
     "parameters": {
+      "max_tokens": 32768,
       "temperature": 1,
       "top_p": 1,
       "stream": true,
-      "max_tokens": 32768,
       "frequency_penalty": 0,
       "presence_penalty": 0
     },
@@ -124,9 +126,9 @@
     "format": "api",
     "settings": {},
     "parameters": {
+      "max_tokens": 65536,
       "temperature": 1,
       "top_p": 1,
-      "max_tokens": 65536,
       "stream": true,
       "frequency_penalty": 0,
       "presence_penalty": 0

extensions/inference-openai-extension/src/index.ts

@@ -74,6 +74,11 @@ export default class JanInferenceOpenAIExtension extends RemoteOAIEngine {
    * @returns
    */
   transformPayload = (payload: OpenAIPayloadType): OpenAIPayloadType => {
+    // Remove empty stop words
+    if (payload.stop?.length === 0) {
+      const { stop, ...params } = payload
+      payload = params
+    }
     // Transform the payload for preview models
     if (this.previewModels.includes(payload.model)) {
       const { max_tokens, stop, ...params } = payload

joi/package.json

@@ -5,7 +5,6 @@
   "module": "dist/esm/index.js",
   "types": "dist/index.d.ts",
   "description": "A collection of UI component",
-  "private": true,
   "files": [
     "dist"
   ],

package.json

@@ -27,12 +27,12 @@
     "pre-install:linux": "find extensions -type f -path \"**/*.tgz\" -exec cp {} pre-install \\;",
     "pre-install:win32": "powershell -Command \"Get-ChildItem -Path \"extensions\" -Recurse -File -Filter \"*.tgz\" | ForEach-Object { Copy-Item -Path $_.FullName -Destination \"pre-install\" }\"",
     "pre-install": "run-script-os",
-    "copy:assets": "cpx \"pre-install/*.tgz\" \"electron/pre-install/\" && cpx \"themes/**\" \"electron/themes\" && cpx \"docs/openapi/**\" \"electron/docs/openapi\"",
+    "copy:assets": "cpx \"pre-install/*.tgz\" \"electron/pre-install/\" && cpx \"themes/**\" \"electron/themes\"",
     "dev:electron": "yarn copy:assets && yarn workspace jan dev",
     "dev:web": "yarn workspace @janhq/web dev",
-    "dev:server": "yarn copy:assets && yarn workspace @janhq/server dev",
+    "dev:server": "yarn workspace @janhq/server dev",
     "dev": "turbo run dev --parallel --filter=!@janhq/server",
-    "build:server": "yarn copy:assets && cd server && yarn install && yarn run build",
+    "build:server": "cd server && yarn install && yarn run build",
     "build:core": "cd core && yarn install && yarn run build",
     "build:web": "yarn workspace @janhq/web build && cpx \"web/out/**\" \"electron/renderer/\"",
     "build:electron": "yarn copy:assets && yarn workspace jan build",

server/index.ts

@@ -1,9 +1,9 @@
 import fastify from 'fastify'
 import dotenv from 'dotenv'
-import { v1Router, log, getJanExtensionsPath } from '@janhq/core/node'
-import { join } from 'path'
+import { log } from '@janhq/core/node'
 import tcpPortUsed from 'tcp-port-used'
 import { Logger } from './helpers/logger'
+import CORTEX_SCHEMA from './cortex.json'
 
 // Load environment variables
 dotenv.config()
@@ -66,34 +66,29 @@ export const startServer = async (configs?: ServerConfig): Promise<boolean> => {
 
     // Initialize Fastify server with logging
     server = fastify({
-      logger: new Logger(),
+      loggerInstance: new Logger(),
       // Set body limit to 100MB - Default is 1MB
       // According to OpenAI - a batch input file can be up to 100 MB in size
       // Whisper endpoints accept up to 25MB
       // Vision endpoints accept up to 4MB
-      bodyLimit: 104_857_600
+      bodyLimit: 104_857_600,
     })
 
     // Register CORS if enabled
     if (corsEnabled) await server.register(require('@fastify/cors'), {})
 
+    CORTEX_SCHEMA.servers[0].url = configs?.prefix ?? '/v1'
     // Register Swagger for API documentation
     await server.register(require('@fastify/swagger'), {
       mode: 'static',
       specification: {
-        path: configs?.schemaPath ?? './../docs/openapi/jan.yaml',
-        baseDir: configs?.baseDir ?? './../docs/openapi',
-        postProcessor: function (swaggerObject: any) {
-          swaggerObject.servers[0].url = configs?.prefix ?? '/v1'
-          return swaggerObject
-        },
+        document: CORTEX_SCHEMA,
       },
     })
 
     // Register Swagger UI
     await server.register(require('@fastify/swagger-ui'), {
       routePrefix: '/',
-      baseDir: configs?.baseDir ?? join(__dirname, '../..', './docs/openapi'),
       uiConfig: {
         docExpansion: 'full',
         deepLinking: false,
@@ -102,26 +97,12 @@ export const startServer = async (configs?: ServerConfig): Promise<boolean> => {
       transformSpecificationClone: true,
     })
 
-    // Register static file serving for extensions
-    // TODO: Watch extension files changes and reload
-    await server.register(
-      (childContext: any, _: any, done: any) => {
-        childContext.register(require('@fastify/static'), {
-          root: getJanExtensionsPath(),
-          wildcard: false,
-        })
+    server.register(require('@fastify/http-proxy'), {
+      upstream: 'http://127.0.0.1:39291/v1',
+      prefix: configs?.prefix ?? '/v1',
+      http2: false,
+    })
 
-        done()
-      },
-      { prefix: 'extensions' }
-    )
-
-    // Register proxy middleware
-    if (configs?.storageAdataper)
-      server.addHook('preHandler', configs.storageAdataper)
-
-    // Register API routes
-    await server.register(v1Router, { prefix: configs?.prefix ?? '/v1' })
     // Start listening for requests
     await server
       .listen({

server/main.ts

@@ -1,7 +1,6 @@
-import { s3 } from './middleware/s3'
 import { setup } from './helpers/setup'
 import { startServer as start } from './index'
 /**
  * Setup extensions and start the server
  */
-setup().then(() => start({ storageAdataper: s3 }))
+setup().then(() => start())

server/middleware/s3.ts

@@ -1,70 +0,0 @@
-import { join } from 'path'
-
-// Middleware to intercept requests and proxy if certain conditions are met
-const config = {
-  endpoint: process.env.AWS_ENDPOINT,
-  region: process.env.AWS_REGION,
-  credentials: {
-    accessKeyId: process.env.AWS_ACCESS_KEY_ID,
-    secretAccessKey: process.env.AWS_SECRET_ACCESS_KEY,
-  },
-}
-
-const S3_BUCKET_NAME = process.env.S3_BUCKET_NAME
-
-const fs = require('@cyclic.sh/s3fs')(S3_BUCKET_NAME, config)
-const PROXY_PREFIX = '/v1/fs'
-const PROXY_ROUTES = ['/threads', '/messages']
-
-export const s3 = (req: any, reply: any, done: any) => {
-  // Proxy FS requests to S3 using S3FS
-  if (req.url.startsWith(PROXY_PREFIX)) {
-    const route = req.url.split('/').pop()
-    const args = parseRequestArgs(req)
-
-    // Proxy matched requests to the s3fs module
-    if (args.length && PROXY_ROUTES.some((route) => args[0].includes(route))) {
-      try {
-        // Handle customized route
-        // S3FS does not handle appendFileSync
-        if (route === 'appendFileSync') {
-          let result = handAppendFileSync(args)
-
-          reply.status(200).send(result)
-          return
-        }
-        // Reroute the other requests to the s3fs module
-        const result = fs[route](...args)
-        reply.status(200).send(result)
-        return
-      } catch (ex) {
-        console.error(ex)
-      }
-    }
-  }
-  // Let other requests go through
-  done()
-}
-
-const parseRequestArgs = (req: Request) => {
-  const {
-    getJanDataFolderPath,
-    normalizeFilePath,
-  } = require('@janhq/core/node')
-
-  return JSON.parse(req.body as any).map((arg: any) =>
-    typeof arg === 'string' &&
-    (arg.startsWith(`file:/`) || arg.startsWith(`file:\\`))
-      ? join(getJanDataFolderPath(), normalizeFilePath(arg))
-      : arg
-  )
-}
-
-const handAppendFileSync = (args: any[]) => {
-  if (fs.existsSync(args[0])) {
-    const data = fs.readFileSync(args[0], 'utf-8')
-    return fs.writeFileSync(args[0], data + args[1])
-  } else {
-    return fs.writeFileSync(args[0], args[1])
-  }
-}

server/package.json

@@ -8,7 +8,8 @@
   "homepage": "https://jan.ai",
   "description": "Use offline LLMs with your own data. Run open source models like Llama2 or Falcon on your internal computers/servers.",
   "files": [
-    "build/**"
+    "build/**",
+    "cortex.json"
   ],
   "scripts": {
     "lint": "eslint . --ext \".js,.jsx,.ts,.tsx\"",
@@ -19,14 +20,15 @@
   "dependencies": {
     "@alumna/reflect": "^1.1.3",
     "@cyclic.sh/s3fs": "^1.2.9",
-    "@fastify/cors": "^8.4.2",
+    "@fastify/cors": "^10.0.1",
+    "@fastify/http-proxy": "^10.0.0",
     "@fastify/static": "^6.12.0",
-    "@fastify/swagger": "^8.13.0",
-    "@fastify/swagger-ui": "2.0.1",
+    "@fastify/swagger": "^9.4.0",
+    "@fastify/swagger-ui": "5.2.0",
     "@janhq/core": "link:./core",
     "@npmcli/arborist": "^7.3.1",
     "dotenv": "^16.3.1",
-    "fastify": "^4.24.3",
+    "fastify": "^5.2.0",
     "fetch-retry": "^5.0.6",
     "node-fetch": "2",
     "request": "^2.88.2",

server/tsconfig.json

@@ -15,7 +15,8 @@
     "paths": { "*": ["node_modules/*"] },
     "typeRoots": ["node_modules/@types"],
     "ignoreDeprecations": "5.0",
-    "declaration": true
+    "declaration": true,
+    "resolveJsonModule": true
   },
   // "sourceMap": true,
 

web/containers/CenterPanelContainer/index.tsx

@@ -1,15 +1,45 @@
 import { PropsWithChildren } from 'react'
 
+import { useMediaQuery } from '@janhq/joi'
 import { useAtomValue } from 'jotai'
 
 import { twMerge } from 'tailwind-merge'
 
+import { MainViewState } from '@/constants/screens'
+
+import { LEFT_PANEL_WIDTH } from '../LeftPanelContainer'
+
+import { RIGHT_PANEL_WIDTH } from '../RightPanelContainer'
+
+import {
+  mainViewStateAtom,
+  showLeftPanelAtom,
+  showRightPanelAtom,
+} from '@/helpers/atoms/App.atom'
 import { reduceTransparentAtom } from '@/helpers/atoms/Setting.atom'
 
-const CenterPanelContainer = ({ children }: PropsWithChildren) => {
+type Props = {
+  isShowStarterScreen?: boolean
+} & PropsWithChildren
+
+const CenterPanelContainer = ({ children, isShowStarterScreen }: Props) => {
   const reduceTransparent = useAtomValue(reduceTransparentAtom)
+  const matches = useMediaQuery('(max-width: 880px)')
+  const showLeftPanel = useAtomValue(showLeftPanelAtom)
+  const showRightPanel = useAtomValue(showRightPanelAtom)
+  const mainViewState = useAtomValue(mainViewStateAtom)
+
   return (
-    <div className={twMerge('flex h-full w-full')}>
+    <div
+      className={twMerge('flex h-full w-full')}
+      style={{
+        maxWidth: matches
+          ? '100%'
+          : mainViewState === MainViewState.Thread && !isShowStarterScreen
+            ? `calc(100% - (${showRightPanel ? Number(localStorage.getItem(RIGHT_PANEL_WIDTH)) : 0}px + ${showLeftPanel ? Number(localStorage.getItem(LEFT_PANEL_WIDTH)) : 0}px))`
+            : '100%',
+      }}
+    >
       <div
         className={twMerge(
           'h-full w-full overflow-hidden bg-[hsla(var(--center-panel-bg))]',

web/containers/ErrorMessage/index.test.tsx

@@ -30,20 +30,23 @@ describe('ErrorMessage Component', () => {
 
   beforeEach(() => {
     jest.clearAllMocks()
-    ;(useAtomValue as jest.Mock).mockReturnValue([])
-    ;(useSetAtom as jest.Mock).mockReturnValue(mockSetMainState)
-    ;(useSetAtom as jest.Mock).mockReturnValue(mockSetSelectedSettingScreen)
-    ;(useSetAtom as jest.Mock).mockReturnValue(mockSetModalTroubleShooting)
-    ;(useSendChatMessage as jest.Mock).mockReturnValue({
-      resendChatMessage: mockResendChatMessage,
-    })
+      ; (useAtomValue as jest.Mock).mockReturnValue([])
+      ; (useSetAtom as jest.Mock).mockReturnValue(mockSetMainState)
+      ; (useSetAtom as jest.Mock).mockReturnValue(mockSetSelectedSettingScreen)
+      ; (useSetAtom as jest.Mock).mockReturnValue(mockSetModalTroubleShooting)
+      ; (useSendChatMessage as jest.Mock).mockReturnValue({
+        resendChatMessage: mockResendChatMessage,
+      })
   })
 
   it('renders error message with InvalidApiKey correctly', () => {
     const message: ThreadMessage = {
       id: '1',
-      status: MessageStatus.Error,
-      error_code: ErrorCode.InvalidApiKey,
+      metadata: {
+        error: MessageStatus.Error,
+        error_code: ErrorCode.InvalidApiKey,
+      },
+      status: "completed",
       content: [{ text: { value: 'Invalid API Key' } }],
     } as ThreadMessage
 
@@ -56,8 +59,11 @@ describe('ErrorMessage Component', () => {
   it('renders general error message correctly', () => {
     const message: ThreadMessage = {
       id: '1',
-      status: MessageStatus.Error,
-      error_code: ErrorCode.Unknown,
+      status: "completed",
+      metadata: {
+        error: MessageStatus.Error,
+        error_code: ErrorCode.Unknown
+      },
       content: [{ text: { value: 'Unknown error occurred' } }],
     } as ThreadMessage
 
@@ -69,9 +75,11 @@ describe('ErrorMessage Component', () => {
   it('opens troubleshooting modal when link is clicked', () => {
     const message: ThreadMessage = {
       id: '1',
-      status: MessageStatus.Error,
-      error_code: ErrorCode.Unknown,
-      content: [{ text: { value: 'Unknown error occurred' } }],
+      status: "completed",
+      metadata: {
+        error: MessageStatus.Error,
+        error_code: ErrorCode.Unknown,
+      }, content: [{ text: { value: 'Unknown error occurred' } }],
     } as ThreadMessage
 
     render(<ErrorMessage message={message} />)

web/containers/ErrorMessage/index.tsx

@@ -14,41 +14,65 @@ import ModalTroubleShooting, {
 
 import { MainViewState } from '@/constants/screens'
 
+import { isLocalEngine } from '@/utils/modelEngine'
+
 import { mainViewStateAtom } from '@/helpers/atoms/App.atom'
 
+import { activeAssistantAtom } from '@/helpers/atoms/Assistant.atom'
 import { selectedSettingAtom } from '@/helpers/atoms/Setting.atom'
-import { activeThreadAtom } from '@/helpers/atoms/Thread.atom'
 
 const ErrorMessage = ({ message }: { message: ThreadMessage }) => {
   const setModalTroubleShooting = useSetAtom(modalTroubleShootingAtom)
   const setMainState = useSetAtom(mainViewStateAtom)
   const setSelectedSettingScreen = useSetAtom(selectedSettingAtom)
-  const activeThread = useAtomValue(activeThreadAtom)
+  const activeAssistant = useAtomValue(activeAssistantAtom)
+
+  const defaultDesc = () => {
+    return (
+      <>
+        <p>
+          {`Something's wrong.`} Access&nbsp;
+          <span
+            className="cursor-pointer text-[hsla(var(--app-link))] underline"
+            onClick={() => setModalTroubleShooting(true)}
+          >
+            troubleshooting assistance
+          </span>
+          &nbsp;now.
+        </p>
+        <ModalTroubleShooting />
+      </>
+    )
+  }
+
+  const getEngine = () => {
+    const engineName = activeAssistant?.model?.engine
+    return engineName ? EngineManager.instance().get(engineName) : null
+  }
 
   const getErrorTitle = () => {
-    switch (message.error_code) {
+    const engine = getEngine()
+
+    switch (message.metadata?.error_code) {
       case ErrorCode.InvalidApiKey:
       case ErrorCode.AuthenticationError:
         return (
-          <span data-testid="invalid-API-key-error">
-            Invalid API key. Please check your API key from{' '}
-            <button
-              className="font-medium text-[hsla(var(--app-link))] underline"
-              onClick={() => {
-                setMainState(MainViewState.Settings)
-
-                if (activeThread?.assistants[0]?.model.engine) {
-                  const engine = EngineManager.instance().get(
-                    activeThread.assistants[0].model.engine
-                  )
+          <>
+            <span data-testid="invalid-API-key-error">
+              Invalid API key. Please check your API key from{' '}
+              <button
+                className="font-medium text-[hsla(var(--app-link))] underline"
+                onClick={() => {
+                  setMainState(MainViewState.Settings)
                   engine?.name && setSelectedSettingScreen(engine.name)
-                }
-              }}
-            >
-              Settings
-            </button>{' '}
-            and try again.
-          </span>
+                }}
+              >
+                Settings
+              </button>{' '}
+              and try again.
+            </span>
+            {defaultDesc()}
+          </>
         )
       default:
         return (
@@ -56,8 +80,20 @@ const ErrorMessage = ({ message }: { message: ThreadMessage }) => {
             data-testid="passthrough-error-message"
             className="first-letter:uppercase"
           >
-            {message.content[0]?.text?.value && (
-              <AutoLink text={message.content[0].text.value} />
+            {message.content[0]?.text?.value === 'Failed to fetch' &&
+            engine &&
+            !isLocalEngine(String(engine?.name)) ? (
+              <span>
+                No internet connection. <br /> Switch to an on-device model or
+                check connection.
+              </span>
+            ) : (
+              <>
+                {message?.content[0]?.text?.value && (
+                  <AutoLink text={message?.content[0]?.text?.value} />
+                )}
+                {defaultDesc()}
+              </>
             )}
           </p>
         )
@@ -65,24 +101,13 @@ const ErrorMessage = ({ message }: { message: ThreadMessage }) => {
   }
 
   return (
-    <div className="mx-auto mt-10 max-w-[700px]">
-      {message.status === MessageStatus.Error && (
+    <div className="mx-auto my-6 max-w-[700px]">
+      {!!message.metadata?.error && (
         <div
           key={message.id}
           className="mx-6 flex flex-col items-center space-y-2 text-center font-medium text-[hsla(var(--text-secondary))]"
         >
           {getErrorTitle()}
-          <p>
-            {`Something's wrong.`} Access&nbsp;
-            <span
-              className="cursor-pointer text-[hsla(var(--app-link))] underline"
-              onClick={() => setModalTroubleShooting(true)}
-            >
-              troubleshooting assistance
-            </span>
-            &nbsp;now.
-          </p>
-          <ModalTroubleShooting />
         </div>
       )}
     </div>

web/containers/Layout/BottomPanel/index.tsx

@@ -35,7 +35,7 @@ const BottomPanel = () => {
   return (
     <div
       className={twMerge(
-        'fixed bottom-0 left-0 z-50 flex h-9 w-full items-center justify-between px-3 text-xs',
+        'fixed bottom-0 left-0 z-40 flex h-9 w-full items-center justify-between px-3 text-xs',
         reduceTransparent &&
           'border-t border-[hsla(var(--app-border))] bg-[hsla(var(--bottom-panel-bg))]'
       )}

web/containers/Layout/TopPanel/index.tsx

@@ -144,7 +144,7 @@ const TopPanel = () => {
             theme="icon"
             onClick={() => {
               setMainViewState(MainViewState.Settings)
-              setSelectedSetting('Appearance')
+              setSelectedSetting('Preferences')
             }}
           >
             <PaletteIcon size={16} className="cursor-pointer" />

web/containers/Layout/index.tsx

@@ -1,9 +1,11 @@
 'use client'
 
-import { useEffect, useMemo } from 'react'
+import { useEffect, useState } from 'react'
 
-import { useAtomValue, useSetAtom } from 'jotai'
+import { Button } from '@janhq/joi'
+import { useAtom, useAtomValue, useSetAtom } from 'jotai'
 
+import posthog from 'posthog-js'
 import { twMerge } from 'tailwind-merge'
 
 import BottomPanel from '@/containers/Layout/BottomPanel'
@@ -31,12 +33,72 @@ import MainViewContainer from '../MainViewContainer'
 import InstallingExtensionModal from './BottomPanel/InstallingExtension/InstallingExtensionModal'
 
 import { mainViewStateAtom } from '@/helpers/atoms/App.atom'
-import { reduceTransparentAtom } from '@/helpers/atoms/Setting.atom'
+import {
+  productAnalyticAtom,
+  productAnalyticPromptAtom,
+  reduceTransparentAtom,
+} from '@/helpers/atoms/Setting.atom'
 
 const BaseLayout = () => {
   const setMainViewState = useSetAtom(mainViewStateAtom)
   const importModelStage = useAtomValue(getImportModelStageAtom)
   const reduceTransparent = useAtomValue(reduceTransparentAtom)
+  const [productAnalytic, setProductAnalytic] = useAtom(productAnalyticAtom)
+  const [productAnalyticPrompt, setProductAnalyticPrompt] = useAtom(
+    productAnalyticPromptAtom
+  )
+  const [showProductAnalyticPrompt, setShowProductAnalyticPrompt] =
+    useState(false)
+
+  useEffect(() => {
+    const timer = setTimeout(() => {
+      if (productAnalyticPrompt) {
+        setShowProductAnalyticPrompt(true)
+      }
+      return () => clearTimeout(timer)
+    }, 3000) // 3 seconds delay
+
+    return () => clearTimeout(timer) // Cleanup timer on unmount
+  }, [productAnalyticPrompt])
+
+  useEffect(() => {
+    if (productAnalytic) {
+      posthog.init(POSTHOG_KEY, {
+        api_host: POSTHOG_HOST,
+        autocapture: false,
+        capture_pageview: false,
+        capture_pageleave: false,
+        disable_session_recording: true,
+        person_profiles: 'always',
+        persistence: 'localStorage',
+        opt_out_capturing_by_default: true,
+        // eslint-disable-next-line @typescript-eslint/naming-convention
+        sanitize_properties: function (properties) {
+          const denylist = [
+            '$pathname',
+            '$initial_pathname',
+            '$current_url',
+            '$initial_current_url',
+            '$host',
+            '$initial_host',
+            '$initial_person_info',
+          ]
+
+          denylist.forEach((key) => {
+            if (properties[key]) {
+              properties[key] = null // Set each denied property to null
+            }
+          })
+
+          return properties
+        },
+      })
+      posthog.opt_in_capturing()
+      posthog.register({ app_version: VERSION })
+    } else {
+      posthog.opt_out_capturing()
+    }
+  }, [productAnalytic])
 
   useEffect(() => {
     if (localStorage.getItem(SUCCESS_SET_NEW_DESTINATION) === 'true') {
@@ -54,6 +116,17 @@ const BaseLayout = () => {
     )
   }, [setMainViewState])
 
+  const handleProductAnalytics = (isAllowed: boolean) => {
+    setProductAnalytic(isAllowed)
+    setProductAnalyticPrompt(false)
+    setShowProductAnalyticPrompt(false)
+    if (isAllowed) {
+      posthog.opt_in_capturing()
+    } else {
+      posthog.opt_out_capturing()
+    }
+  }
+
   return (
     <div
       className={twMerge(
@@ -76,6 +149,79 @@ const BaseLayout = () => {
         <ChooseWhatToImportModal />
         <InstallingExtensionModal />
         <HuggingFaceRepoDetailModal />
+        {showProductAnalyticPrompt && (
+          <div className="fixed bottom-4 z-50 m-4 max-w-full rounded-xl border border-[hsla(var(--app-border))] bg-[hsla(var(--app-bg))] p-6 shadow-2xl sm:bottom-8 sm:right-4 sm:m-0 sm:max-w-[400px]">
+            <div className="mb-4 flex items-center gap-x-2">
+              <svg
+                width="32"
+                height="32"
+                viewBox="0 0 32 32"
+                fill="none"
+                xmlns="http://www.w3.org/2000/svg"
+              >
+                <path
+                  d="M5.5 12.5C5.5 11.1193 6.61929 10 8 10H24C25.3807 10 26.5 11.1193 26.5 12.5V18.5C26.5 24.299 21.799 29 16 29C10.201 29 5.5 24.299 5.5 18.5V12.5Z"
+                  fill="#2563EB"
+                />
+                <path
+                  fillRule="evenodd"
+                  clipRule="evenodd"
+                  d="M8.20959 25.54L12.0093 10H14.0093L9.84984 27.0113C9.25274 26.579 8.70292 26.0855 8.20959 25.54ZM11.5993 28.0361C11.2955 27.8957 10.9996 27.7412 10.7124 27.5734L15.0093 10H16.0093L11.5993 28.0361Z"
+                  fill="white"
+                />
+                <path
+                  d="M21 8C21 6.67392 20.4732 5.40215 19.5355 4.46447C18.5979 3.52678 17.3261 3 16 3C14.6739 3 13.4021 3.52678 12.4645 4.46447C11.5268 5.40215 11 6.67392 11 8"
+                  stroke="#2563EB"
+                  strokeWidth="2"
+                  strokeLinecap="round"
+                />
+                <path
+                  d="M27.0478 18.054C27.609 18.5733 27.609 19.4267 27.0478 19.946C25.221 21.6363 20.9622 25 16 25C11.0378 25 6.77899 21.6363 4.95219 19.946C4.39099 19.4267 4.39099 18.5733 4.95219 18.054C6.77899 16.3637 11.0378 13 16 13C20.9622 13 25.221 16.3637 27.0478 18.054Z"
+                  fill="#C8D1EA"
+                />
+                <circle cx="16" cy="19" r="4" fill="#2563EB" />
+                <path
+                  d="M19.25 17.5C19.9404 17.5 20.5 16.9404 20.5 16.25C20.5 15.5596 19.9404 15 19.25 15C18.5596 15 18 15.5596 18 16.25C18 16.9404 18.5596 17.5 19.25 17.5Z"
+                  fill="white"
+                />
+                <path
+                  d="M17.75 18.5C18.1642 18.5 18.5 18.1642 18.5 17.75C18.5 17.3358 18.1642 17 17.75 17C17.3358 17 17 17.3358 17 17.75C17 18.1642 17.3358 18.5 17.75 18.5Z"
+                  fill="white"
+                />
+              </svg>
+
+              <h6 className="text-base font-semibold">Help Us Improve Jan</h6>
+            </div>
+            <p className="text-[hsla(var(--text-secondary))]">
+              To improve Jan, we collect anonymous data to understand feature
+              usage. Your chats and personal information are never tracked. You
+              can change this anytime in&nbsp;
+              <span className="font-semibold">{`Settings > Privacy.`}</span>
+            </p>
+            <p className="mt-6 text-[hsla(var(--text-secondary))]">
+              Would you like to help us to improve Jan?
+            </p>
+            <div className="mt-6 flex items-center gap-x-2">
+              <Button
+                onClick={() => {
+                  handleProductAnalytics(true)
+                }}
+              >
+                Allow
+              </Button>
+              <Button
+                data-testid="btn-deny-product-analytics"
+                theme="ghost"
+                variant="outline"
+                onClick={() => {
+                  handleProductAnalytics(false)
+                }}
+              >
+                Deny
+              </Button>
+            </div>
+          </div>
+        )}
       </div>
       <BottomPanel />
     </div>

web/containers/LeftPanelContainer/index.tsx

@@ -17,7 +17,7 @@ import { reduceTransparentAtom } from '@/helpers/atoms/Setting.atom'
 type Props = PropsWithChildren
 
 const DEFAULT_LEFT_PANEL_WIDTH = 200
-const LEFT_PANEL_WIDTH = 'leftPanelWidth'
+export const LEFT_PANEL_WIDTH = 'leftPanelWidth'
 
 const LeftPanelContainer = ({ children }: Props) => {
   const [leftPanelRef, setLeftPanelRef] = useState<HTMLDivElement | null>(null)
@@ -106,7 +106,7 @@ const LeftPanelContainer = ({ children }: Props) => {
           <Fragment>
             <div
               className={twMerge(
-                'group/resize absolute right-0 top-0 z-[9999] h-full w-1 flex-shrink-0 flex-grow-0 resize-x blur-sm hover:cursor-col-resize hover:bg-[hsla(var(--resize-bg))]',
+                'group/resize absolute right-0 top-0 z-40 h-full w-1 flex-shrink-0 flex-grow-0 resize-x blur-sm hover:cursor-col-resize hover:bg-[hsla(var(--resize-bg))]',
                 isResizing && 'cursor-col-resize bg-[hsla(var(--resize-bg))]',
                 !reduceTransparent && 'shadow-sm'
               )}

web/containers/MainViewContainer/index.tsx

@@ -37,7 +37,7 @@ const MainViewContainer = () => {
   }
 
   return (
-    <div className={twMerge('relative flex w-full')}>
+    <div className={twMerge('relative flex w-[calc(100%-48px)]')}>
       <div className="w-full">
         <m.div
           key={mainViewState}

web/containers/ModelDropdown/index.tsx

@@ -12,7 +12,7 @@ import {
   useClickOutside,
 } from '@janhq/joi'
 
-import { useAtom, useAtomValue, useSetAtom } from 'jotai'
+import { atom, useAtom, useAtomValue, useSetAtom } from 'jotai'
 
 import {
   ChevronDownIcon,
@@ -37,6 +37,7 @@ import useUpdateModelParameters from '@/hooks/useUpdateModelParameters'
 
 import { formatDownloadPercentage, toGibibytes } from '@/utils/converter'
 
+import { manualRecommendationModel } from '@/utils/model'
 import {
   getLogoEngine,
   getTitleByEngine,
@@ -46,6 +47,7 @@ import {
 
 import { extensionManager } from '@/extension'
 
+import { activeAssistantAtom } from '@/helpers/atoms/Assistant.atom'
 import { inActiveEngineProviderAtom } from '@/helpers/atoms/Extension.atom'
 import {
   configuredModelsAtom,
@@ -64,17 +66,23 @@ type Props = {
   disabled?: boolean
 }
 
+export const modelDropdownStateAtom = atom(false)
+
 const ModelDropdown = ({
   disabled,
   chatInputMode,
   strictedThread = true,
 }: Props) => {
   const { downloadModel } = useDownloadModel()
+  const [modelDropdownState, setModelDropdownState] = useAtom(
+    modelDropdownStateAtom
+  )
 
   const [searchFilter, setSearchFilter] = useState('local')
   const [searchText, setSearchText] = useState('')
-  const [open, setOpen] = useState(false)
+  const [open, setOpen] = useState<boolean>(modelDropdownState)
   const activeThread = useAtomValue(activeThreadAtom)
+  const activeAssistant = useAtomValue(activeAssistantAtom)
   const downloadingModels = useAtomValue(getDownloadingModelAtom)
   const [toggle, setToggle] = useState<HTMLDivElement | null>(null)
   const [selectedModel, setSelectedModel] = useAtom(selectedModelAtom)
@@ -82,22 +90,38 @@ const ModelDropdown = ({
   const [dropdownOptions, setDropdownOptions] = useState<HTMLDivElement | null>(
     null
   )
+
   const downloadStates = useAtomValue(modelDownloadStateAtom)
   const setThreadModelParams = useSetAtom(setThreadModelParamsAtom)
   const { updateModelParameter } = useUpdateModelParameters()
   const searchInputRef = useRef<HTMLInputElement>(null)
   const configuredModels = useAtomValue(configuredModelsAtom)
-  const featuredModel = configuredModels.filter((x) =>
-    x.metadata?.tags?.includes('Featured')
+
+  const featuredModel = configuredModels.filter(
+    (x) =>
+      manualRecommendationModel.includes(x.id) &&
+      x.metadata?.tags?.includes('Featured') &&
+      x.metadata?.size < 5000000000
   )
   const { updateThreadMetadata } = useCreateNewThread()
 
-  useClickOutside(() => setOpen(false), null, [dropdownOptions, toggle])
+  useClickOutside(() => handleChangeStateOpen(false), null, [
+    dropdownOptions,
+    toggle,
+  ])
 
   const [showEngineListModel, setShowEngineListModel] = useAtom(
     showEngineListModelAtom
   )
 
+  const handleChangeStateOpen = useCallback(
+    (state: boolean) => {
+      setOpen(state)
+      setModelDropdownState(state)
+    },
+    [setModelDropdownState]
+  )
+
   const isModelSupportRagAndTools = useCallback((model: Model) => {
     return (
       model?.engine === InferenceEngine.openai ||
@@ -143,6 +167,12 @@ const ModelDropdown = ({
     [configuredModels, searchText, searchFilter, downloadedModels]
   )
 
+  useEffect(() => {
+    if (modelDropdownState && chatInputMode) {
+      setOpen(modelDropdownState)
+    }
+  }, [chatInputMode, modelDropdownState])
+
   useEffect(() => {
     if (open && searchInputRef.current) {
       searchInputRef.current.focus()
@@ -151,17 +181,24 @@ const ModelDropdown = ({
 
   useEffect(() => {
     if (!activeThread) return
-    const modelId = activeThread?.assistants?.[0]?.model?.id
+    const modelId = activeAssistant?.model?.id
 
     let model = downloadedModels.find((model) => model.id === modelId)
     if (!model) {
-      model = recommendedModel
+      model = undefined
     }
     setSelectedModel(model)
-  }, [recommendedModel, activeThread, downloadedModels, setSelectedModel])
+  }, [
+    recommendedModel,
+    activeThread,
+    downloadedModels,
+    setSelectedModel,
+    activeAssistant?.model?.id,
+  ])
 
   const onClickModelItem = useCallback(
     async (modelId: string) => {
+      if (!activeAssistant) return
       const model = downloadedModels.find((m) => m.id === modelId)
       setSelectedModel(model)
       setOpen(false)
@@ -172,14 +209,14 @@ const ModelDropdown = ({
           ...activeThread,
           assistants: [
             {
-              ...activeThread.assistants[0],
+              ...activeAssistant,
               tools: [
                 {
                   type: 'retrieval',
                   enabled: isModelSupportRagAndTools(model as Model),
                   settings: {
-                    ...(activeThread.assistants[0].tools &&
-                      activeThread.assistants[0].tools[0]?.settings),
+                    ...(activeAssistant.tools &&
+                      activeAssistant.tools[0]?.settings),
                   },
                 },
               ],
@@ -192,8 +229,12 @@ const ModelDropdown = ({
           model?.settings.ctx_len ?? 8192
         )
         const overriddenParameters = {
-          ctx_len: Math.min(8192, model?.settings.ctx_len ?? 8192),
-          max_tokens: defaultContextLength,
+          ctx_len: !isLocalEngine(model?.engine)
+            ? undefined
+            : defaultContextLength,
+          max_tokens: !isLocalEngine(model?.engine)
+            ? (model?.parameters.max_tokens ?? 8192)
+            : defaultContextLength,
         }
 
         const modelParams = {
@@ -215,13 +256,14 @@ const ModelDropdown = ({
       }
     },
     [
+      activeAssistant,
       downloadedModels,
-      activeThread,
       setSelectedModel,
+      activeThread,
+      updateThreadMetadata,
       isModelSupportRagAndTools,
       setThreadModelParams,
       updateModelParameter,
-      updateThreadMetadata,
     ]
   )
 
@@ -329,14 +371,21 @@ const ModelDropdown = ({
               'inline-block max-w-[200px] cursor-pointer overflow-hidden text-ellipsis whitespace-nowrap',
               open && 'border border-transparent'
             )}
-            onClick={() => setOpen(!open)}
+            onClick={() => handleChangeStateOpen(!open)}
           >
-            <span>{selectedModel?.name}</span>
+            <span
+              className={twMerge(
+                !selectedModel && 'text-[hsla(var(--text-tertiary))]'
+              )}
+            >
+              {selectedModel?.name || 'Select Model'}
+            </span>
           </Badge>
         ) : (
           <Input
             value={selectedModel?.name || ''}
             className="cursor-pointer"
+            placeholder="Select Model"
             disabled={disabled}
             readOnly
             suffixIcon={

web/containers/Providers/Jotai.tsx

@@ -4,9 +4,11 @@ import { PropsWithChildren } from 'react'
 
 import { Provider, atom } from 'jotai'
 
+import { FileInfo } from '@/types/file'
+
 export const editPromptAtom = atom<string>('')
 export const currentPromptAtom = atom<string>('')
-export const fileUploadAtom = atom<FileInfo[]>([])
+export const fileUploadAtom = atom<FileInfo | undefined>()
 
 export const searchAtom = atom<string>('')
 
@@ -15,10 +17,3 @@ export const selectedTextAtom = atom('')
 export default function JotaiWrapper({ children }: PropsWithChildren) {
   return <Provider>{children}</Provider>
 }
-
-export type FileType = 'image' | 'pdf'
-
-export type FileInfo = {
-  file: File
-  type: FileType
-}

web/containers/Providers/ModelHandler.tsx

@@ -31,6 +31,7 @@ import {
   addNewMessageAtom,
   updateMessageAtom,
   tokenSpeedAtom,
+  deleteMessageAtom,
 } from '@/helpers/atoms/ChatMessage.atom'
 import { downloadedModelsAtom } from '@/helpers/atoms/Model.atom'
 import {
@@ -49,6 +50,7 @@ export default function ModelHandler() {
   const addNewMessage = useSetAtom(addNewMessageAtom)
   const updateMessage = useSetAtom(updateMessageAtom)
   const downloadedModels = useAtomValue(downloadedModelsAtom)
+  const deleteMessage = useSetAtom(deleteMessageAtom)
   const activeModel = useAtomValue(activeModelAtom)
   const setActiveModel = useSetAtom(activeModelAtom)
   const setStateModel = useSetAtom(stateModelAtom)
@@ -86,7 +88,7 @@ export default function ModelHandler() {
   }, [activeModelParams])
 
   const onNewMessageResponse = useCallback(
-    (message: ThreadMessage) => {
+    async (message: ThreadMessage) => {
       if (message.type === MessageRequestType.Thread) {
         addNewMessage(message)
       }
@@ -154,12 +156,15 @@ export default function ModelHandler() {
         ...thread,
 
         title: cleanedMessageContent,
-        metadata: thread.metadata,
+        metadata: {
+          ...thread.metadata,
+          title: cleanedMessageContent,
+        },
       }
 
       extensionManager
         .get<ConversationalExtension>(ExtensionTypeEnum.Conversational)
-        ?.saveThread({
+        ?.modifyThread({
           ...updatedThread,
         })
         .then(() => {
@@ -233,7 +238,9 @@ export default function ModelHandler() {
 
       const thread = threadsRef.current?.find((e) => e.id == message.thread_id)
       if (!thread) return
+
       const messageContent = message.content[0]?.text?.value
+
       const metadata = {
         ...thread.metadata,
         ...(messageContent && { lastMessage: messageContent }),
@@ -246,15 +253,31 @@ export default function ModelHandler() {
 
       extensionManager
         .get<ConversationalExtension>(ExtensionTypeEnum.Conversational)
-        ?.saveThread({
+        ?.modifyThread({
           ...thread,
           metadata,
         })
 
-      // If this is not the summary of the Thread, don't need to add it to the Thread
-      extensionManager
-        .get<ConversationalExtension>(ExtensionTypeEnum.Conversational)
-        ?.addNewMessage(message)
+      if (message.status === MessageStatus.Error) {
+        message.metadata = {
+          ...message.metadata,
+          error: message.content[0]?.text?.value,
+          error_code: message.error_code,
+        }
+      }
+      ;(async () => {
+        const updatedMessage = await extensionManager
+          .get<ConversationalExtension>(ExtensionTypeEnum.Conversational)
+          ?.createMessage(message)
+          .catch(() => undefined)
+        if (updatedMessage) {
+          deleteMessage(message.id)
+          addNewMessage(updatedMessage)
+          setTokenSpeed((prev) =>
+            prev ? { ...prev, message: updatedMessage.id } : undefined
+          )
+        }
+      })()
 
       // Attempt to generate the title of the Thread when needed
       generateThreadTitle(message, thread)
@@ -279,7 +302,9 @@ export default function ModelHandler() {
 
   const generateThreadTitle = (message: ThreadMessage, thread: Thread) => {
     // If this is the first ever prompt in the thread
-    if (thread.title?.trim() !== defaultThreadTitle) {
+    if (
+      (thread.title ?? thread.metadata?.title)?.trim() !== defaultThreadTitle
+    ) {
       return
     }
 
@@ -292,11 +317,14 @@ export default function ModelHandler() {
       const updatedThread: Thread = {
         ...thread,
         title: (thread.metadata?.lastMessage as string) || defaultThreadTitle,
-        metadata: thread.metadata,
+        metadata: {
+          ...thread.metadata,
+          title: (thread.metadata?.lastMessage as string) || defaultThreadTitle,
+        },
       }
       return extensionManager
         .get<ConversationalExtension>(ExtensionTypeEnum.Conversational)
-        ?.saveThread({
+        ?.modifyThread({
           ...updatedThread,
         })
         .then(() => {
@@ -313,7 +341,7 @@ export default function ModelHandler() {
 
     if (!threadMessages || threadMessages.length === 0) return
 
-    const summarizeFirstPrompt = `Summarize in a ${maxWordForThreadTitle}-word Title. Give the title only. "${threadMessages[0].content[0].text.value}"`
+    const summarizeFirstPrompt = `Summarize in a ${maxWordForThreadTitle}-word Title. Give the title only. "${threadMessages[0]?.content[0]?.text?.value}"`
 
     // Prompt: Given this query from user {query}, return to me the summary in 10 words as the title
     const msgId = ulid()
@@ -330,6 +358,7 @@ export default function ModelHandler() {
       id: msgId,
       threadId: message.thread_id,
       type: MessageRequestType.Summary,
+      attachments: [],
       messages,
       model: {
         ...activeModelRef.current,

web/containers/Providers/Responsive.tsx

@@ -11,15 +11,14 @@ const Responsive = () => {
   const [showRightPanel, setShowRightPanel] = useAtom(showRightPanelAtom)
 
   // Refs to store the last known state of the panels
-  const lastLeftPanelState = useRef<boolean>(true)
-  const lastRightPanelState = useRef<boolean>(true)
+  const lastLeftPanelState = useRef<boolean>(showLeftPanel)
+  const lastRightPanelState = useRef<boolean>(showRightPanel)
 
   useEffect(() => {
     if (matches) {
       // Store the last known state before closing the panels
       lastLeftPanelState.current = showLeftPanel
       lastRightPanelState.current = showRightPanel
-
       setShowLeftPanel(false)
       setShowRightPanel(false)
     } else {

web/containers/RightPanelContainer/index.tsx

@@ -16,13 +16,13 @@ import { reduceTransparentAtom } from '@/helpers/atoms/Setting.atom'
 
 type Props = PropsWithChildren
 
-const DEFAULT_RIGTH_PANEL_WIDTH = 280
-const RIGHT_PANEL_WIDTH = 'rightPanelWidth'
+const DEFAULT_RIGHT_PANEL_WIDTH = 280
+export const RIGHT_PANEL_WIDTH = 'rightPanelWidth'
 
 const RightPanelContainer = ({ children }: Props) => {
   const [isResizing, setIsResizing] = useState(false)
   const [threadRightPanelWidth, setRightPanelWidth] = useState(
-    Number(localStorage.getItem(RIGHT_PANEL_WIDTH)) || DEFAULT_RIGTH_PANEL_WIDTH
+    Number(localStorage.getItem(RIGHT_PANEL_WIDTH)) || DEFAULT_RIGHT_PANEL_WIDTH
   )
   const [rightPanelRef, setRightPanelRef] = useState<HTMLDivElement | null>(
     null
@@ -55,11 +55,11 @@ const RightPanelContainer = ({ children }: Props) => {
               mouseMoveEvent.clientX <
             200
           ) {
-            setRightPanelWidth(DEFAULT_RIGTH_PANEL_WIDTH)
+            setRightPanelWidth(DEFAULT_RIGHT_PANEL_WIDTH)
             setIsResizing(false)
             localStorage.setItem(
               RIGHT_PANEL_WIDTH,
-              String(DEFAULT_RIGTH_PANEL_WIDTH)
+              String(DEFAULT_RIGHT_PANEL_WIDTH)
             )
             setShowRightPanel(false)
           } else {
@@ -77,8 +77,8 @@ const RightPanelContainer = ({ children }: Props) => {
 
   useEffect(() => {
     if (localStorage.getItem(RIGHT_PANEL_WIDTH) === null) {
-      setRightPanelWidth(DEFAULT_RIGTH_PANEL_WIDTH)
-      localStorage.setItem(RIGHT_PANEL_WIDTH, String(DEFAULT_RIGTH_PANEL_WIDTH))
+      setRightPanelWidth(DEFAULT_RIGHT_PANEL_WIDTH)
+      localStorage.setItem(RIGHT_PANEL_WIDTH, String(DEFAULT_RIGHT_PANEL_WIDTH))
     }
     window.addEventListener('mousemove', resize)
     window.addEventListener('mouseup', stopResizing)
@@ -109,7 +109,7 @@ const RightPanelContainer = ({ children }: Props) => {
           <Fragment>
             <div
               className={twMerge(
-                'group/resize absolute left-0 top-0 z-[9999] h-full w-1 flex-shrink-0 flex-grow-0 resize-x blur-sm hover:cursor-col-resize hover:bg-[hsla(var(--resize-bg))]',
+                'group/resize absolute left-0 top-0 z-40 h-full w-1 flex-shrink-0 flex-grow-0 resize-x blur-sm hover:cursor-col-resize hover:bg-[hsla(var(--resize-bg))]',
                 isResizing && 'cursor-col-resize bg-[hsla(var(--resize-bg))]',
                 !reduceTransparent && 'shadow-sm'
               )}

web/helpers/atoms/App.atom.ts

@@ -1,14 +1,25 @@
 import { atom } from 'jotai'
 
+import { atomWithStorage } from 'jotai/utils'
+
 import { MainViewState } from '@/constants/screens'
 
 export const mainViewStateAtom = atom<MainViewState>(MainViewState.Thread)
 
 export const defaultJanDataFolderAtom = atom<string>('')
 
+const SHOW_RIGHT_PANEL = 'showRightPanel'
+
 // Store panel atom
 export const showLeftPanelAtom = atom<boolean>(true)
-export const showRightPanelAtom = atom<boolean>(true)
+
+export const showRightPanelAtom = atomWithStorage<boolean>(
+  SHOW_RIGHT_PANEL,
+  false,
+  undefined,
+  { getOnInit: true }
+)
+
 export const showSystemMonitorPanelAtom = atom<boolean>(false)
 export const appDownloadProgressAtom = atom<number>(-1)
 export const updateVersionErrorAtom = atom<string | undefined>(undefined)

web/helpers/atoms/Assistant.atom.ts

@@ -1,4 +1,12 @@
-import { Assistant } from '@janhq/core'
+import { Assistant, ThreadAssistantInfo } from '@janhq/core'
 import { atom } from 'jotai'
+import { atomWithStorage } from 'jotai/utils'
 
 export const assistantsAtom = atom<Assistant[]>([])
+
+/**
+ * Get the current active assistant
+ */
+export const activeAssistantAtom = atomWithStorage<
+  ThreadAssistantInfo | undefined
+>('activeAssistant', undefined, undefined, { getOnInit: true })

web/helpers/atoms/ChatMessage.atom.ts

@@ -6,6 +6,8 @@ import {
 } from '@janhq/core'
 import { atom } from 'jotai'
 
+import { atomWithStorage } from 'jotai/utils'
+
 import {
   getActiveThreadIdAtom,
   updateThreadStateLastMessageAtom,
@@ -13,15 +15,32 @@ import {
 
 import { TokenSpeed } from '@/types/token'
 
+const CHAT_MESSAGE_NAME = 'chatMessages'
 /**
  * Stores all chat messages for all threads
  */
-export const chatMessages = atom<Record<string, ThreadMessage[]>>({})
+export const chatMessagesStorage = atomWithStorage<
+  Record<string, ThreadMessage[]>
+>(CHAT_MESSAGE_NAME, {}, undefined, { getOnInit: true })
+
+export const cachedMessages = atom<Record<string, ThreadMessage[]>>()
+/**
+ * Retrieve chat messages for all threads
+ */
+export const chatMessages = atom(
+  (get) => get(cachedMessages) ?? get(chatMessagesStorage),
+  (_get, set, newValue: Record<string, ThreadMessage[]>) => {
+    set(cachedMessages, newValue)
+    ;(() => set(chatMessagesStorage, newValue))()
+  }
+)
 
 /**
  * Stores the status of the messages load for each thread
  */
-export const readyThreadsMessagesAtom = atom<Record<string, boolean>>({})
+export const readyThreadsMessagesAtom = atomWithStorage<
+  Record<string, boolean>
+>('currentThreadMessages', {}, undefined, { getOnInit: true })
 
 /**
  * Store the token speed for current message
@@ -34,6 +53,7 @@ export const getCurrentChatMessagesAtom = atom<ThreadMessage[]>((get) => {
   const activeThreadId = get(getActiveThreadIdAtom)
   if (!activeThreadId) return []
   const messages = get(chatMessages)[activeThreadId]
+  if (!Array.isArray(messages)) return []
   return messages ?? []
 })
 
@@ -121,7 +141,7 @@ export const deleteMessageAtom = atom(null, (get, set, id: string) => {
   if (threadId) {
     // Should also delete error messages to clear out the error state
     newData[threadId] = newData[threadId].filter(
-      (e) => e.id !== id && e.status !== MessageStatus.Error
+      (e) => e.id !== id && !e.metadata?.error
     )
 
     set(chatMessages, newData)

web/helpers/atoms/Model.atom.test.ts

@@ -58,7 +58,9 @@ describe('Model.atom.ts', () => {
         setAtom.current({ id: '1' } as any)
       })
       expect(getAtom.current).toEqual([{ id: '1' }])
-      reset.current([])
+      act(() => {
+        reset.current([])
+      })
     })
   })
 
@@ -83,7 +85,9 @@ describe('Model.atom.ts', () => {
         removeAtom.current('1')
       })
       expect(getAtom.current).toEqual([])
-      reset.current([])
+      act(() => {
+        reset.current([])
+      })
     })
   })
 
@@ -113,7 +117,9 @@ describe('Model.atom.ts', () => {
         removeAtom.current('1')
       })
       expect(getAtom.current).toEqual([])
-      reset.current([])
+      act(() => {
+        reset.current([])
+      })
     })
   })
 

web/helpers/atoms/Setting.atom.ts

@@ -11,9 +11,12 @@ export const janSettingScreenAtom = atom<SettingScreen[]>([])
 export const THEME = 'themeAppearance'
 export const REDUCE_TRANSPARENT = 'reduceTransparent'
 export const SPELL_CHECKING = 'spellChecking'
+export const PRODUCT_ANALYTIC = 'productAnalytic'
+export const PRODUCT_ANALYTIC_PROMPT = 'productAnalyticPrompt'
 export const THEME_DATA = 'themeData'
 export const THEME_OPTIONS = 'themeOptions'
 export const THEME_PATH = 'themePath'
+export const CHAT_WIDTH = 'chatWidth'
 export const themesOptionsAtom = atomWithStorage<
   { name: string; value: string }[]
 >(THEME_OPTIONS, [], undefined, { getOnInit: true })
@@ -47,3 +50,21 @@ export const spellCheckAtom = atomWithStorage<boolean>(
   undefined,
   { getOnInit: true }
 )
+export const productAnalyticAtom = atomWithStorage<boolean>(
+  PRODUCT_ANALYTIC,
+  false,
+  undefined,
+  { getOnInit: true }
+)
+export const productAnalyticPromptAtom = atomWithStorage<boolean>(
+  PRODUCT_ANALYTIC_PROMPT,
+  true,
+  undefined,
+  { getOnInit: true }
+)
+export const chatWidthAtom = atomWithStorage<string>(
+  CHAT_WIDTH,
+  'full',
+  undefined,
+  { getOnInit: true }
+)

web/helpers/atoms/Thread.atom.ts

@@ -207,7 +207,7 @@ export const setThreadModelParamsAtom = atom(
  */
 export const activeSettingInputBoxAtom = atomWithStorage<boolean>(
   ACTIVE_SETTING_INPUT_BOX,
-  false,
+  true,
   undefined,
   { getOnInit: true }
 )

web/hooks/useActiveModel.ts

@@ -8,8 +8,8 @@ import { toaster } from '@/containers/Toast'
 import { LAST_USED_MODEL_ID } from './useRecommendedModel'
 
 import { vulkanEnabledAtom } from '@/helpers/atoms/AppConfig.atom'
+import { activeAssistantAtom } from '@/helpers/atoms/Assistant.atom'
 import { downloadedModelsAtom } from '@/helpers/atoms/Model.atom'
-import { activeThreadAtom } from '@/helpers/atoms/Thread.atom'
 
 export const activeModelAtom = atom<Model | undefined>(undefined)
 export const loadModelErrorAtom = atom<string | undefined>(undefined)
@@ -28,12 +28,12 @@ export const stateModelAtom = atom<ModelState>({
 
 export function useActiveModel() {
   const [activeModel, setActiveModel] = useAtom(activeModelAtom)
-  const activeThread = useAtomValue(activeThreadAtom)
   const [stateModel, setStateModel] = useAtom(stateModelAtom)
   const downloadedModels = useAtomValue(downloadedModelsAtom)
   const setLoadModelError = useSetAtom(loadModelErrorAtom)
   const pendingModelLoad = useRef(false)
   const isVulkanEnabled = useAtomValue(vulkanEnabledAtom)
+  const activeAssistant = useAtomValue(activeAssistantAtom)
 
   const downloadedModelsRef = useRef<Model[]>([])
 
@@ -79,12 +79,12 @@ export function useActiveModel() {
     }
 
     /// Apply thread model settings
-    if (activeThread?.assistants[0]?.model.id === modelId) {
+    if (activeAssistant?.model.id === modelId) {
       model = {
         ...model,
         settings: {
           ...model.settings,
-          ...activeThread.assistants[0].model.settings,
+          ...activeAssistant?.model.settings,
         },
       }
     }

web/hooks/useCreateNewThread.test.ts

@@ -67,7 +67,7 @@ describe('useCreateNewThread', () => {
       } as any)
     })
 
-    expect(mockSetAtom).toHaveBeenCalledTimes(6) // Check if all the necessary atoms were set
+    expect(mockSetAtom).toHaveBeenCalledTimes(1)
     expect(extensionManager.get).toHaveBeenCalled()
   })
 
@@ -104,7 +104,7 @@ describe('useCreateNewThread', () => {
       await result.current.requestCreateNewThread({
         id: 'assistant1',
         name: 'Assistant 1',
-        instructions: "Hello Jan Assistant",
+        instructions: 'Hello Jan Assistant',
         model: {
           id: 'model1',
           parameters: [],
@@ -113,16 +113,8 @@ describe('useCreateNewThread', () => {
       } as any)
     })
 
-    expect(mockSetAtom).toHaveBeenCalledTimes(6) // Check if all the necessary atoms were set
+    expect(mockSetAtom).toHaveBeenCalledTimes(1) // Check if all the necessary atoms were set
     expect(extensionManager.get).toHaveBeenCalled()
-    expect(mockSetAtom).toHaveBeenNthCalledWith(
-      2,
-      expect.objectContaining({
-        assistants: expect.arrayContaining([
-          expect.objectContaining({ instructions: 'Hello Jan Assistant' }),
-        ]),
-      })
-    )
   })
 
   it('should create a new thread with previous instructions', async () => {
@@ -166,16 +158,8 @@ describe('useCreateNewThread', () => {
       } as any)
     })
 
-    expect(mockSetAtom).toHaveBeenCalledTimes(6) // Check if all the necessary atoms were set
+    expect(mockSetAtom).toHaveBeenCalledTimes(1) // Check if all the necessary atoms were set
     expect(extensionManager.get).toHaveBeenCalled()
-    expect(mockSetAtom).toHaveBeenNthCalledWith(
-      2,
-      expect.objectContaining({
-        assistants: expect.arrayContaining([
-          expect.objectContaining({ instructions: 'Hello Jan' }),
-        ]),
-      })
-    )
   })
 
   it('should show a warning toast if trying to create an empty thread', async () => {
@@ -212,13 +196,12 @@ describe('useCreateNewThread', () => {
 
     const { result } = renderHook(() => useCreateNewThread())
 
-    const mockThread = { id: 'thread1', title: 'Test Thread' }
+    const mockThread = { id: 'thread1', title: 'Test Thread', assistants: [{}] }
 
     await act(async () => {
       await result.current.updateThreadMetadata(mockThread as any)
     })
 
     expect(mockUpdateThread).toHaveBeenCalledWith(mockThread)
-    expect(extensionManager.get).toHaveBeenCalled()
   })
 })

web/hooks/useCreateNewThread.ts

@@ -1,7 +1,6 @@
 import { useCallback } from 'react'
 
 import {
-  Assistant,
   ConversationalExtension,
   ExtensionTypeEnum,
   Thread,
@@ -9,24 +8,28 @@ import {
   ThreadState,
   AssistantTool,
   Model,
+  Assistant,
 } from '@janhq/core'
-import { atom, useAtomValue, useSetAtom } from 'jotai'
+import { atom, useAtom, useAtomValue, useSetAtom } from 'jotai'
+
+import { useDebouncedCallback } from 'use-debounce'
 
 import { copyOverInstructionEnabledAtom } from '@/containers/CopyInstruction'
 import { fileUploadAtom } from '@/containers/Providers/Jotai'
 
 import { toaster } from '@/containers/Toast'
 
-import { generateThreadId } from '@/utils/thread'
+import { isLocalEngine } from '@/utils/modelEngine'
 
 import { useActiveModel } from './useActiveModel'
-import useRecommendedModel from './useRecommendedModel'
 
+import useRecommendedModel from './useRecommendedModel'
 import useSetActiveThread from './useSetActiveThread'
 
 import { extensionManager } from '@/extension'
 
 import { experimentalFeatureEnabledAtom } from '@/helpers/atoms/AppConfig.atom'
+import { activeAssistantAtom } from '@/helpers/atoms/Assistant.atom'
 import { selectedModelAtom } from '@/helpers/atoms/Model.atom'
 import {
   threadsAtom,
@@ -34,7 +37,6 @@ import {
   updateThreadAtom,
   setThreadModelParamsAtom,
   isGeneratingResponseAtom,
-  activeThreadAtom,
 } from '@/helpers/atoms/Thread.atom'
 
 const createNewThreadAtom = atom(null, (get, set, newThread: Thread) => {
@@ -64,25 +66,25 @@ export const useCreateNewThread = () => {
   const copyOverInstructionEnabled = useAtomValue(
     copyOverInstructionEnabledAtom
   )
-  const activeThread = useAtomValue(activeThreadAtom)
+  const [activeAssistant, setActiveAssistant] = useAtom(activeAssistantAtom)
 
   const experimentalEnabled = useAtomValue(experimentalFeatureEnabledAtom)
   const setIsGeneratingResponse = useSetAtom(isGeneratingResponseAtom)
 
-  const { recommendedModel, downloadedModels } = useRecommendedModel()
-
   const threads = useAtomValue(threadsAtom)
   const { stopInference } = useActiveModel()
 
+  const { recommendedModel } = useRecommendedModel()
+
   const requestCreateNewThread = async (
-    assistant: Assistant,
+    assistant: (ThreadAssistantInfo & { id: string; name: string }) | Assistant,
     model?: Model | undefined
   ) => {
     // Stop generating if any
     setIsGeneratingResponse(false)
     stopInference()
 
-    const defaultModel = model ?? recommendedModel ?? downloadedModels[0]
+    const defaultModel = model || recommendedModel
 
     if (!model) {
       // if we have model, which means user wants to create new thread from Model hub. Allow them.
@@ -113,18 +115,22 @@ export const useCreateNewThread = () => {
     )
 
     const overriddenSettings = {
-      ctx_len: defaultContextLength,
+      ctx_len: !isLocalEngine(defaultModel?.engine)
+        ? undefined
+        : defaultContextLength,
     }
 
     // Use ctx length by default
     const overriddenParameters = {
-      max_tokens: defaultContextLength,
+      max_tokens: !isLocalEngine(defaultModel?.engine)
+        ? (defaultModel?.parameters.token_limit ?? 8192)
+        : defaultContextLength,
     }
 
     const createdAt = Date.now()
     let instructions: string | undefined = assistant.instructions
     if (copyOverInstructionEnabled) {
-      instructions = activeThread?.assistants[0]?.instructions ?? undefined
+      instructions = activeAssistant?.instructions ?? undefined
     }
     const assistantInfo: ThreadAssistantInfo = {
       assistant_id: assistant.id,
@@ -139,46 +145,97 @@ export const useCreateNewThread = () => {
       instructions,
     }
 
-    const threadId = generateThreadId(assistant.id)
-    const thread: Thread = {
-      id: threadId,
+    const thread: Partial<Thread> = {
       object: 'thread',
       title: 'New Thread',
       assistants: [assistantInfo],
       created: createdAt,
       updated: createdAt,
+      metadata: {
+        title: 'New Thread',
+      },
     }
 
     // add the new thread on top of the thread list to the state
     //TODO: Why do we have thread list then thread states? Should combine them
-    createNewThread(thread)
+    try {
+      const createdThread = await persistNewThread(thread, assistantInfo)
+      if (!createdThread) throw 'Thread created failed.'
+      createNewThread(createdThread)
 
-    setSelectedModel(defaultModel)
-    setThreadModelParams(thread.id, {
-      ...defaultModel?.settings,
-      ...defaultModel?.parameters,
-      ...overriddenSettings,
-    })
+      setSelectedModel(defaultModel)
+      setThreadModelParams(createdThread.id, {
+        ...defaultModel?.settings,
+        ...defaultModel?.parameters,
+        ...overriddenSettings,
+      })
 
-    // Delete the file upload state
-    setFileUpload([])
-    // Update thread metadata
-    await updateThreadMetadata(thread)
-
-    setActiveThread(thread)
+      // Delete the file upload state
+      setFileUpload(undefined)
+      setActiveThread(createdThread)
+    } catch (ex) {
+      return toaster({
+        title: 'Thread created failed.',
+        description: `To avoid piling up empty threads, please reuse previous one before creating new.`,
+        type: 'error',
+      })
+    }
   }
 
+  const updateThreadExtension = (thread: Thread) => {
+    return extensionManager
+      .get<ConversationalExtension>(ExtensionTypeEnum.Conversational)
+      ?.modifyThread(thread)
+  }
+
+  const updateAssistantExtension = (
+    threadId: string,
+    assistant: ThreadAssistantInfo
+  ) => {
+    return extensionManager
+      .get<ConversationalExtension>(ExtensionTypeEnum.Conversational)
+      ?.modifyThreadAssistant(threadId, assistant)
+  }
+
+  const updateThreadCallback = useDebouncedCallback(updateThreadExtension, 300)
+  const updateAssistantCallback = useDebouncedCallback(
+    updateAssistantExtension,
+    300
+  )
+
   const updateThreadMetadata = useCallback(
     async (thread: Thread) => {
       updateThread(thread)
 
-      await extensionManager
-        .get<ConversationalExtension>(ExtensionTypeEnum.Conversational)
-        ?.saveThread(thread)
+      setActiveAssistant(thread.assistants[0])
+      updateThreadCallback(thread)
+      updateAssistantCallback(thread.id, thread.assistants[0])
     },
-    [updateThread]
+    [
+      updateThread,
+      setActiveAssistant,
+      updateThreadCallback,
+      updateAssistantCallback,
+    ]
   )
 
+  const persistNewThread = async (
+    thread: Partial<Thread>,
+    assistantInfo: ThreadAssistantInfo
+  ): Promise<Thread | undefined> => {
+    return await extensionManager
+      .get<ConversationalExtension>(ExtensionTypeEnum.Conversational)
+      ?.createThread(thread)
+      .then(async (thread) => {
+        await extensionManager
+          .get<ConversationalExtension>(ExtensionTypeEnum.Conversational)
+          ?.createThreadAssistant(thread.id, assistantInfo)
+          .catch(console.error)
+        return thread
+      })
+      .catch(() => undefined)
+  }
+
   return {
     requestCreateNewThread,
     updateThreadMetadata,