# UseTested
> Receipts-verified reviews of agent-economy services (x402 endpoints, MCP servers, AI APIs). Every review is anchored to on-chain purchase receipts — no receipt, no review.

## Endpoints
- GET /v1/search?q=<query>&category=<inference|data|search|media|infrastructure>: free — service directory with verdicts, scores, price per call.
- GET /v1/review?service=<name>: free — full review JSON: verdict, scores, measured latency/error rate, receipts[] (on-chain tx proofs), methodology_version.
- GET /v1/bundle?category=<category>: $0.05 — all reviews in a category, sorted by reliability. The one paid route (x402, USDC on Base): bulk convenience, and settlement keeps our Bazaar indexing trigger alive.

## MCP server
- POST /mcp — streamable-http MCP server (stateless; use the official MCP SDK StreamableHTTP client). Tools: search_services (free), get_review (free), get_bundle ($0.05 x402). Paid tools settle via x402 (USDC on Base) — challenge arrives as the JSON-RPC error envelope; echo the payment payload in params._meta["x402/payment"].

## Usage
Before spending an agent budget on a paid service, read its review to check aliveness, real price vs advertised, latency and error rate. Search and single reviews are free to read; attribution appreciated.

## Contact
usetested@redd.in