Docs /

Endpoints

• POST https://mcp.answerline.dev/mcp with Authorization: Bearer <key>.
• POST https://mcp.answerline.dev/<API_KEY>/mcp for clients that cannot send headers. The URL contains your key: treat it as a secret.

The server speaks Streamable HTTP over POST and supports protocol revisions 2024-11-05 through 2025-11-25 and 2026-07-28. Clients that accept text/event-stream get tool calls streamed with keep-alive comments, so long calls survive proxies. GET and DELETE answer 405. Browser requests from origins not on the server's allow list are refused with 403.

Connecting a client

claude mcp add --transport http answerline https://mcp.answerline.dev/mcp --header "Authorization: Bearer $API_KEY"

{
  "mcpServers": {
    "answerline": {
      "url": "https://mcp.answerline.dev/mcp",
      "headers": {
        "Authorization": "Bearer sk_..."
      }
    }
  }
}

https://mcp.answerline.dev/<API_KEY>/mcp

• Cursor: add to .cursor/mcp.json (or any client that reads an mcpServers JSON file)
• Claude Desktop: Settings, Connectors, Add custom connector, then paste this URL with your key in it

For clients that only take a URL, use the path form with your key in it.

Tools

One tool per engine your account can use, plus credits and engines. Each engine tool's input schema is that endpoint's request schema, so its arguments are the request body listed in the API reference.

A tool result is the API's response. API errors come back as results marked isError with the error body, so the model can read its code. Charge, balance and rate-limit headers and the request id are in the result's _meta.

Costs and timeouts

An engine tool call is a synchronous API call: it costs the engine's price, the same as an async task, and can take up to ten minutes, so raise your client's tool timeout where it is configurable. For recurring or large runs, use async tasks and batches instead.