Single tool 'rh' with critical definition gaps. Tool name lacks action verb (should be 'call_robinhood_endpoint' or similar). Description is 97 chars but vague, does not explain WHEN to use it, what it returns, or prerequisites. Input schema present but severely underspecified: 'pay' parameter has no type (function is not a JSON Schema type), 'body' is generic object with no field documentation, 'endpoint' lacks enum of valid values or format constraints. No output schema documented. No error handling guidance. Tool appears to be a generic pass-through wrapper rather than a well-defined agent interface.

• Rename tool to 'call_robinhood_endpoint' or 'query_robinhood_chain' to signal action verb and improve LLM discoverability.
• Replace 'pay' parameter with server-side secret injection (environment variable or vault). If client must provide a signer, document it as a configuration requirement, not a tool parameter.
• Convert 'endpoint' to an enum: ['rhchain/balance', 'rhchain/tx', 'rhchain/registry', 'rhchain/watch', 'ai/chat']. Add per-endpoint parameter descriptions (e.g., 'endpoint=rhchain/balance requires address field in body').
• Document output schema for each endpoint. Example: 'rhchain/balance returns {usdg_balance: number, native_balance: number, address: string}'. Include pagination guidance if applicable.
• Expand tool description to 150-250 chars: 'Call any Robinhood Chain endpoint (balance, transaction, registry, watch) or AI inference endpoint. Requires x402 payment wrapper. Returns endpoint-specific JSON. Use for blockchain queries, token lookups, and LLM inference on Robinhood Chain.'