forecast-accuracy-mcp (Python)

Python version of the forecast-accuracy-mcp stdio MCP server, built on the official mcp Python SDK. Same tools, same APIFY_TOKEN env var, same hosted endpoint as the Node.js server.

Stdio MCP server: forecast track records: forecasts vs actuals, earnings beats, prediction markets, archived news via Apify's hosted MCP gateway

It answers initialize and tools/list locally from the bundled tools.json (no token, no network). tools/call is forwarded to Apify's hosted MCP endpoint (mcp.apify.com) with your own APIFY_TOKEN, so each run uses your Apify account. Without a token a call returns a clear error instead of running anything. The token is only sent in the Authorization header and is never logged.

Install

Requires Python 3.10+. With uv:

uvx --from "git+https://github.com/GeminiGeorge22/forecast-accuracy-mcp#subdirectory=python" forecast-accuracy-mcp

With pip:

pip install "git+https://github.com/GeminiGeorge22/forecast-accuracy-mcp#subdirectory=python"
forecast-accuracy-mcp

MCP client config:

{
  "mcpServers": {
    "forecast-accuracy": {
      "command": "uvx",
      "args": ["--from", "git+https://github.com/GeminiGeorge22/forecast-accuracy-mcp#subdirectory=python", "forecast-accuracy-mcp"],
      "env": { "APIFY_TOKEN": "YOUR_APIFY_TOKEN" }
    }
  }
}

Optional env: APIFY_ACTORS (comma-separated subset of the Actors) and APIFY_MCP_URL (full endpoint override).

Tests

Tests use an in-process mock server, with no network and no token. They fail if APIFY_TOKEN is set.

cd python && pip install -e ".[test]"
env -u APIFY_TOKEN python -m pytest

python scripts/build_tools.py --check compares the bundled tools.json with each Actor's live public input schema and exits 1 on any mismatch.