Tool call examples

Endpoint

https://evalscopebench.clauxel.com/mcp

Authentication

Production calls require a paid bearer token. The checkout and token-claim endpoints return machine-readable instructions for agents.

Available tools

• run_benchmark_gate returns structured JSON with verdict, reason, receipt_id, usage_units, and next_action.
• compare_model_scores returns structured JSON with verdict, reason, receipt_id, usage_units, and next_action.
• read_benchmark_report returns structured JSON with verdict, reason, receipt_id, usage_units, and next_action.
• issue_benchmark_receipt returns structured JSON with verdict, reason, receipt_id, usage_units, and next_action.

Example call

{"jsonrpc":"2.0","id":"call-1","method":"tools/call","params":{"name":"run_benchmark_gate","arguments":{"sample":"EvalScope Benchmark MCP sample with public-safe workflow context, owner, policy, deadline, risk notes, and reviewer evidence."}}}

Setup pages

• Claude Desktop setup
• Cursor setup
• VS Code setup
• ChatGPT agent setup
• Tool call examples
• Security notes