A paid remote MCP that runs safety evaluations before you ship agent releases. It exposes five tools: run_safety_replay for re-running evals, check_policy_gate for compliance checks, record_eval_receipt for audit trails, suggest_control_fix for remediation guidance, and export_release_evidence for deployment documentation. You authenticate with a bearer token from the Clauxel dashboard and call it over streamable HTTP. Reach for this when you need structured gates and paper trails between your agent builds and production deployments, especially if you're working in regulated environments or need reproducible safety checks that generate export-ready evidence.
claude mcp add --transport http com.clauxel.safetyreplay-safetyreplay-mcp https://safetyreplay.clauxel.com/mcp