Interactive dashboard for Context Forge projects. Visualize initiatives, slices, documents, and maintenance work across your project portfolio.
- Project portfolio view — See all tracked projects at a glance with initiative cards, progress bars, and document counts
- MCP integration — Live data from Context Forge's MCP server; falls back to local parsing when MCP is unavailable
- Collector cards — Aggregated views for maintenance/operations work and future work items across slice plans
- Project management panel — Add, remove, and refresh projects from the UI; auto-discover projects in a directory
- Zero dependencies — Pure Python server, vanilla JS frontend with in-browser JSX transform
# Clone and enter the project
git clone https://github.com/ecorkran/context-visualizer.git
cd context-visualizer
# Start the server (no install required — zero runtime dependencies)
python3 serve.py
# Open http://localhost:5678No virtual environment or pip install step is needed — the server has zero runtime dependencies (uv is only used for dev tooling like tests). This makes python3 serve.py a good default for AI coding agents to launch the dashboard directly, without provisioning a venv first.
Note: on some machines, first startup can take up to ~30s due to a slow reverse-DNS lookup (socket.getfqdn()) inside Python's stdlib HTTPServer. This is a local networking quirk, not a hang — subsequent requests are fast once it's listening.
Create a mcp-config.json in the project root to connect to a Context Forge MCP server:
{
"prefer": "mcp",
"enableFutureWorkCollector": true,
"server": {
"transport": "stdio",
"command": "node",
"args": ["path/to/context-forge/packages/mcp-server/dist/index.js"],
"env": {}
}
}Without MCP, the visualizer uses parse.py to read project structure directly from the filesystem. Add projects via the UI panel or by placing structure JSON files in the projects/ directory.
# Install dev dependencies
uv sync
# Run all tests (excluding E2E)
pytest -m "not e2e"
# Run E2E browser tests (requires playwright)
playwright install chromium
pytest tests/test_ui_smoke.py
# Run all tests
pytest tests/| File | Purpose |
|---|---|
serve.py |
HTTP server with API endpoints and MCP client management |
mcp_client.py |
Minimal stdlib-only MCP client (JSON-RPC over stdio) |
parse.py |
Local filesystem parser for project structure |
index.html |
Frontend entry point with data loading logic |
project-structure-viz.jsx |
React components (in-browser Babel transform) |
MIT