NoContext

Uptime monitoring for public MCP servers — because when your server is down, your AI has no context.

View live status page →

---

How it works

1. GitHub Actions runs scripts/check.py every 15 minutes.
2. The script sends an MCP initialize handshake (JSON-RPC 2.0) to each server.
3. Results are written to docs/data/status.json and committed back to the repo.
4. GitHub Pages serves docs/index.html, which fetches the JSON at runtime.

No external services, no database — just git and GitHub Actions.

Adding a server

Edit config/servers.yml:

servers:
  - name: "My MCP Server"
    url: "https://my-mcp-server.example.com"
    transport: http
    tags: [custom]

Open a PR and it will be included in the next check run.

What counts as "up"?

The checker POSTs an MCP initialize request and expects a valid JSON-RPC 2.0 response (result or error at the top level). Servers that respond with HTTP 401/403 (auth required) are treated as up — they're reachable, just need credentials.

Status	Meaning
🟢 up	Valid MCP response received
🟡 degraded	Reachable but response is malformed
🔴 down	Timeout, connection refused, or HTTP error

Local development

pip install requests pyyaml
python scripts/check.py
# opens docs/index.html in your browser to preview

Setup for your own fork

1. Fork this repo.
2. Go to Settings → Pages → set source to docs/ on main.
3. Go to Settings → Actions → General → enable "Read and write permissions" for the Actions workflow.
4. The first run will populate docs/data/status.json and the status page will go live.

Contributing

PRs to add new public MCP servers are welcome. Please verify the server is publicly accessible before submitting.