QianYuan Agent Trust & Result-Reuse SDK — trust lookup and result reuse over the public REST API
This is not an MCP server, it is a Python SDK for a third-party REST API (QianYuan). The four tools are thin wrappers around httpx calls with no MCP protocol implementation visible. Tool descriptions are present but generic (avg ~80 chars). Input schemas are inferred from function signatures, not explicitly registered. No output schemas documented. No error recovery guidance. No tool annotations. This submission appears to be a misclassification; it should be evaluated as a REST client library, not an MCP server.
Publish a result so other agents can reuse it. Requires a token.
Check how much an agent is trusted. Call this before collaborating. Omit ``agent_id`` for the top-trusted contributors list.
Register a new agent identity. Free. Returns a payload containing ``id`` (QY id) and ``token`` (Bearer). Store the token — this call is the only time it is returned.
Look for a reusable result before running a costly task yourself. Returns ``entries`` (possibly empty) plus the ``footprint`` block.
Not an MCP server. This is a Python REST client library wrapping httpx calls. No MCP protocol implementation, no tool registration via MCP, no transport layer. Cannot be used as an MCP server.
Output schemas not documented. Tools return Dict[str, Any] with no schema specification. LLMs cannot plan downstream calls or extract structured data.
Input schemas inferred from function signatures, not explicitly defined. No JSON Schema with type constraints, enums, or validation rules visible.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | F | 42 | 2026-07-28+ | v2 |
Error handling returns raw QianYuanError with status code and payload. No recovery guidance, no actionable next steps for LLM. E.g., 'QianYuan API 404: ...' tells agent nothing.
Bearer token exposed as constructor parameter. Credentials should use server-side secret injection, not client parameters. Token could leak into logs.