CoinBucha MCP — Bitcoin signal intelligence for AI agents

Signal-first Bitcoin intelligence over MCP. Leading, non-price indicators — sovereign adoption, hiring velocity, network hashrate, ETF flows, corporate treasuries — each as a 0–100 strength, a direction, a one-line why, and primary sources.

Information, not financial advice. No tool returns a buy/sell/hold recommendation.

This is the public MCP interface for CoinBucha. The hosted server reads the same live signal data the dashboard renders, so the machine face and the human face never diverge.

Connect (hosted — recommended, no install)

The server is remote and speaks Streamable HTTP. Point any MCP client at:

https://coinbucha.com/api/mcp

Claude Desktop / Claude Code — add a custom connector with that URL, or in claude_desktop_config.json:

{
  "mcpServers": {
    "coinbucha": {
      "type": "streamable-http",
      "url": "https://coinbucha.com/api/mcp"
    }
  }
}

Cursor / other clients — add an MCP server with transport streamable-http and the URL above.

Free tier is open (IP rate-limited). Higher limits are available with an API key via the Authorization: Bearer <key> header — see coinbucha.com.

Connect (local stdio — offline/dev)

The bundled server.py exposes the same tools and fetches the live public snapshots from coinbucha.com, so it needs no local data pipeline.

pip install -r requirements.txt
python3 server.py            # stdio transport
# inspect: uv run mcp dev server.py

{
  "mcpServers": {
    "coinbucha": { "command": "python3", "args": ["server.py"] }
  }
}

Tools

Resource coinbucha://methodology — how every signal is computed (see docs/methodology.md). Prompt morning_bitcoin_brief — one-shot template to render the Daily Brew from live signals.

Every response carries an as_of timestamp, a why, and a disclaimer.

Example

Call scan_signals with { "min_strength": 60, "direction": "tailwind" } → the strongest structural tailwinds right now, ranked, each with its rationale and sources. Or run the morning_bitcoin_brief prompt for a written digest.

Methodology & guardrail

Signals are leading, non-price indicators rendered as signal_strength (0–100), a direction, and a why. The full, versioned methodology is public — see docs/methodology.md — because an auditable method is the point.

CoinBucha's public and MCP surfaces are self-audited against MiCA / FCA / GDPR. No surface presents a signal as advice.

Links

• Dashboard: https://coinbucha.com
• MCP endpoint: https://coinbucha.com/api/mcp
• Discovery manifest: https://coinbucha.com/.well-known/mcp.json

License

MIT — see LICENSE. This repository is the public MCP interface and methodology; the signal-collection pipeline is maintained separately.