Generate privacy-safe operational reports from agent run ledgers and local Codex or Claude JSONL/session artifacts. Use when reviewing per-run status, measured duration and tokens, source coverage, unavailable fields, or scheduled-agent reliability without exposing prompts, responses, repository paths, or company-specific data.
