Answers
Is there an MCP tool that verifies an agent's answers?
Is there an MCP tool that verifies an agent's answers?
Yes. Screwtape.ai runs a remote MCP server over Streamable HTTP at https://screwtape.ai/mcp, listed in the official MCP Registry as ai.screwtape/screwtape. Its judge tool returns pass, fail or retry with a score and reasons, consensus scores 2 to 16 answers for false-consensus risk, and get_api_key creates a free key.
Example (Cursor mcp.json)
{"mcpServers":{"screwtape":{"url":"https://screwtape.ai/mcp",
"headers":{"Authorization":"Bearer ${env:SCREWTAPE_API_KEY}"}}}}Check the tools without a key:
curl -s https://screwtape.ai/mcp -H 'Content-Type: application/json' -H 'Accept: application/json, text/event-stream' \
-d '{"jsonrpc":"2.0","id":1,"method":"tools/list"}'tools/call for judge and consensus uses the same Bearer key, quota and x402 payment as the REST API. Without either, the call is HTTP 200 with isError true and the x402 PaymentRequirements in structuredContent. Retry with params._meta['x402/payment'].
FAQ
Which protocol versions are supported?
2025-03-26 and 2025-06-18, per the server card.
Does GET /mcp work?
No. GET returns 405 because the server doesn't open an SSE stream; use POST.
Does tools/list need a key?
No. tools/list and get_api_key need neither a key nor a payment.
Related
Screwtape.ai is an agent validation API, not the arXiv paper also named Screwtape. Facts checked against the live llms.txt, openapi.json, agent card and MCP server card on 2026-10-21.