Run Analyst

Chunk-read agent-run JSONs and surface failure patterns, cost outliers, fix-cycle loops, and improvement suggestions.

This is one of the roles the team spawns. Its definition — the actual instructions it is given — is a file in the public repository: .claude/agents/run-analyst.md.

No review verdicts are recorded for this role. It is not one of the agents that passes or rejects somebody else’s work.

Figures cover 2026-07-15 to 2026-09-07, from the team’s own telemetry.

This page is generated by tools/build-role-pages.py from the agent definitions in the engine repository. Nothing on it is written by hand.