Overview

The documents module provides a complete UI for uploading files, managing knowledge bases, viewing documents with chunk highlighting, and attaching knowledge bases to chat sessions.

Component Architecture

Knowledge Base Pages
    │
    ├── KnowledgeBaseList (TanStack Table)
    │   ├── Search + sort + pagination
    │   ├── Create KB modal
    │   └── Delete KB confirmation
    │
    └── KnowledgeBaseDetail
        ├── KB metadata (name, description, access)
        ├── DocumentsTable (documents in this KB)
        └── DocumentUpload (add files to KB)

Document Pages
    │
    ├── WorkspaceDocumentsList
    │   ├── Documents table
    │   └── Upload button
    │
    └── DocumentViewerPage
        ├── PDF renderer (pdfjs-dist)
        ├── Chunk highlighting
        └── Page navigation

Chat Integration
    │
    └── ChatSettingsForm
        ├── Attach/detach KBs to session
        └── KB selector (multi-select command)

Pages

Features

Document Upload

Drag-and-drop or click-to-select file upload using the S3 presigned URL flow. Supports multi-file upload with per-file progress tracking.

Knowledge Base Management

• Create, edit, delete knowledge bases
• Share with workspace or restrict to teams
• Add/remove documents from KBs
• View document count and status per KB

Document Viewer

Full-page PDF viewer powered by pdfjs-dist:

• Page-by-page rendering
• Chunk highlight overlay (when opened from a citation)
• Page navigation controls
• Responsive sizing

Chat Integration

Knowledge bases are attached to chat sessions via the settings panel. Once attached, the AI agent can search documents in those KBs using the search_knowledge_base tool.

Permission-Based UI