How lucasastorian/llmwiki works
LLM Wiki captures files, web clips, highlights, and notes into local or hosted knowledge bases; an AI client can then search and maintain wiki pages, while the web app presents sources, pages, and their relationships. The graph shows the documented capture-to-indexing flow and the separate MCP maintenance path. Hosted processing, converter engines, and several UI/API links are represented at subsystem level where the README or source index supports them; unsampled internals are not inferred.
Main entry points: main.py API application · local_processor.py Local processor · ocr.py OCR and extraction · search.py MCP tools
Components 21
Capture
- Browser clipper
extension/src/entrypoints/content/index.ts - File upload
api/routes/local_upload.py - Workspace watcher
api/domain/watcher.py
Document processing
- Web parser
api/html_parser/parser.py - Local processor
api/domain/local_processor.py - OCR and extraction
api/services/ocr.py - Document converter
converter/extraction_service.py - Text chunker
api/services/chunker.py
Knowledge services
- Hosted services
api/services/hosted.py - Local services
api/services/local.py - Quiz features
api/routes/quiz.py - Knowledge graph
api/services/graph.py
User interfaces
- API application
api/main.py - Wiki web app
web/src/components/kb/KBDetail.tsx - MCP server
mcp/local_server.py - MCP tools
mcp/tools/search.py
Persistence
- Local index
api/infra/db/sqlite.py - MCP vault
mcp/vaultfs/sqlite.py - Object storage
api/services/s3.py
Other components
- User
- AI assistant
Connections 24
- User → Browser clipper: clips pages
- Browser clipper → API application: saves clips
- User → File upload: uploads files
- User → Workspace watcher: adds workspace files
- File upload → Local processor: indexes file
api/routes/local_upload.py - Workspace watcher → Local processor: dispatches processing
api/domain/watcher.py - Local processor → Local index: updates index
api/domain/local_processor.py - Local processor → Text chunker: chunks text
api/domain/local_processor.py - Local processor → OCR and extraction: extracts PDF text
api/domain/local_processor.py - OCR and extraction → Document converter: requests conversion
api/services/ocr.py - API application → Web parser: parses web clips
- User → Wiki web app: browses wiki
- Wiki web app → API application: uses API
- AI assistant → MCP server: connects via MCP
- MCP server → MCP tools: provides tools
mcp/tools/search.py - MCP tools → MCP vault: searches vault
mcp/tools/search.py - AI assistant → MCP tools: reads and updates wiki
- API application → Hosted services: routes hosted requests
api/main.py - API application → Local services: routes local requests
api/main.py - API application → Knowledge graph: serves graph features
- API application → Quiz features: serves quiz features
api/main.py - Hosted services → Object storage: stores source files
- Local services → Local index: uses local index
- API application → OCR and extraction: runs extraction
api/main.py