← ClaudeAtlas

auditlisted

Audit a job's lineage: read metadata, trace, lineage, verdicts. Triggers: audit, lineage, verdict, history.
produtoramaxvision/maxvision-utilities · ★ 0 · AI & Automation · score 66
Install: claude install-skill produtoramaxvision/maxvision-utilities
# media-forge:audit Job audit and lineage skill. Reads all metadata for a given job ID: trace entries, lineage chain, quality verdicts, cost breakdown, and model call log. ## Workflow 1. Receive job ID from $ARGUMENTS. 2. Locate job dir at `$CLAUDE_PROJECT_DIR/.media-forge/jobs/<jobId>/`. 3. Read `trace.jsonl`, `lineage.json`, `refined_spec.json`, and all verdict JSONs. 4. Summarize: agent chain, model calls, retry count, final verdict, total cost. 5. Display structured audit report to user. ## When to use Invoke when user wants to inspect what happened in a previous generation job: which agents ran, what the quality scores were, how many retries occurred, and the total cost. ## Outputs - Agent chain (ordered list of agents dispatched) - Prompt evolution (initial brief → refined spec) - Quality verdicts per attempt - Retry count and root causes - Total cost in USD - Asset paths