How joaofernandesuk/reponary works
Reponary is a local-first browser workspace combining code-graph inspection with Markdown engineering knowledge. The main workflow is to validate and register a source folder, optionally generate its Graphify graph on the host, then browse project status, graph data, notes, and search results. The host agent owns privileged host operations; the web application serves read-only project and vault data. Graph generation and deployment details are described in the README, while some operation implementations were not sampled, so those links are limited to documented behavior.
Main entry points: app.py HTTP API · agent.py Native agent · index.py Search index · knowledge.py Knowledge index
Components 14
Browser interface
- Workspace UI
frontend/src/main.tsx - HTTP client
frontend/src/api.ts
Web application
- HTTP API
backend/app.py - Control bridge
backend/control.py - Knowledge index
backend/knowledge.py - Search index
backend/index.py
Host control
- Native agent
host/agent.py
Project data
- Source folders
- Graphify outputs
- Markdown vault
- SQLite FTS5 cache
Other components
- Local user
- Graphify CLI
- Docker Compose
Connections 15
- Local user → Workspace UI: uses
- Workspace UI → HTTP client: requests
- HTTP client → HTTP API: calls
- HTTP API → Control bridge: submits operations
- Native agent → HTTP API: polls and reports
- HTTP API → Knowledge index: reads notes
- HTTP API → Search index: searches
- Knowledge index → Markdown vault: reads and indexes
- Search index → Knowledge index: refreshes notes
- Search index → Graphify outputs: indexes graph data
- Search index → SQLite FTS5 cache: writes and queries
- Native agent → Source folders: validates and manages
- Native agent → Graphify CLI: generates graphs
- Native agent → Docker Compose: manages web service
- HTTP API → Graphify outputs: reads graph status