Skillmonger — the human-expertise layer for AI

An HTTP API, MCP server, and llms.txt that route high-stakes questions to a verified human expert. Pay per answer in USDC (x402) or card.

Endpoints for your AI agent

HTTP API: https://api.skillmonger.ai
OpenAPI / Swagger: https://api.skillmonger.ai/docs
Agent-readable docs (read first): https://api.skillmonger.ai/llms.txt
Drop-in MCP server (stdio, PyPI): pip install skillmonger-mcp
OAuth-authenticated MCP HTTP transport: https://api.skillmonger.ai/mcp/*

Quickstart (Stripe-on-behalf-of-user)

curl -X POST https://api.skillmonger.ai/requests/submit \
  -H "Content-Type: application/json" \
  -d '{
    "query": "Detailed question with all context the expert needs",
    "tags": ["employment-law", "arbitration"],
    "desired_output": "yes/no plus 2-sentence reasoning",
    "price": 50,
    "delivery_email": "user@example.com"
  }'
# → 200 { request_id, approval_url, expires_at }
# Hand approval_url to the user. They pay; matching starts atomically.
# Poll: GET /requests/{id}?t=<access_token> until status="completed".

Quickstart (x402 / USDC on Base, no account)

# 1) Discovery — no payment, no email → HTTP 402 with USDC requirements
# 2) Sign with your wallet, resubmit with PAYMENT-SIGNATURE header → 200
# See https://api.skillmonger.ai/llms.txt for the exact x402 v2 flow.

→ Full integration reference (llms.txt)