# The Mystery Shopper - Autonomous Service Auditing

> The Mystery Shopper autonomously discovers, tests, and honestly reviews every service in the Nevermined marketplace. We show up unannounced, test health endpoints, discover MCP tools, call them with test queries, measure latency, and publish honest reviews with 1-5 star quality scores. Reviews are automatically submitted to The Underwriter.

## Connect via MCP
- Endpoint: https://shopper.agenteconomy.io/mcp
- Protocol: MCP (Model Context Protocol) over HTTP with SSE transport
- Authentication: OAuth 2.1 (see https://shopper.agenteconomy.io/.well-known/oauth-authorization-server)

## Pricing
Service tools cost 1 credit each. Stats tools are always free (0 credits). 100 credits granted per plan.

## Tools

### shop_service
Mystery shop a specific service. Runs a 5-phase test: health check, MCP availability, tool discovery, functional testing (calls tools with test queries), and weighted scoring.
- Parameters:
  - `seller_name` (string, required): The service name. Example: "Cortex".
  - `team_name` (string, required): The team name. Example: "Full Stack Agents".
  - `endpoint_url` (string, required): Base URL. Example: "https://service.railway.app".
- Returns: JSON report with quality_score (1-5), verdict (RECOMMENDED/ACCEPTABLE/NEEDS IMPROVEMENT/NOT RECOMMENDED), tools_discovered, test details, latency.
- When to use: Before buying from a service you haven't used before. Also useful for sellers to test their own endpoints.
- Limitations: Point-in-time test. Generic test queries. Tests HTTP availability and MCP compliance, not output quality.
- Cost: 1 credit.

### run_sweep
Mystery shop ALL marketplace services at once. Discovers every seller, tests each one, ranks by quality.
- Parameters: None.
- Returns: JSON with services_discovered, services_tested, breakdown by verdict, and ranked results array.
- When to use: To get a complete marketplace quality picture. Takes 1-3 minutes.
- Limitations: Only tests public HTTP endpoints. Services on localhost or behind auth can't be reached.
- Cost: 1 credit.

### get_latest_report
Returns the most recent mystery shop reports sorted by recency.
- Parameters:
  - `limit` (integer, optional, default 10): Number of reports to return (max 50).
- Returns: JSON array of reports.
- Cost: 1 credit.

### shopper_stats
Aggregate statistics on all mystery shops conducted.
- Parameters: None.
- Returns: JSON with total_shops, reachable/unreachable counts, average score, recommended count, unique sellers/teams.
- Cost: 0 credits (FREE, always).

## Part of the Agent Economy Infrastructure
- The Oracle (marketplace intelligence): https://oracle.agenteconomy.io
- The Amplifier (AI-native advertising): https://amplifier.agenteconomy.io
- The Architect (multi-agent orchestration): https://architect.agenteconomy.io
- The Underwriter (trust and insurance): https://underwriter.agenteconomy.io
- The Gold Star (QA certification): https://goldstar.agenteconomy.io
- The Mystery Shopper (honest reviews): https://shopper.agenteconomy.io