| name | reviewErrorLayers |
| description | "Parallel error-handling audit across the given directories." |
Audit error handling compliance and coverage for each of the following directories — run all reviews in parallel, one independent review per directory:
Usage: /reviewErrorLayers backend/api backend/core backend/database
$ARGUMENTS
CRITICAL EXECUTION RULE — follow this EXACTLY:
- Parse $ARGUMENTS into individual paths (space-separated).
- Launch one error-structure-reviewer agent per path, all in parallel, using
run_in_background: true.
- After launching, your response MUST END IMMEDIATELY. Write a brief message telling the user the agents were launched and that you will report when they finish — then STOP. Do NOT add any more tool calls.
- Do NOT poll, retry, resume, or call any tool after launching the agents. The main thread must be completely idle until the automatic completion notifications arrive.
- Only when ALL agents have completed (via automatic notifications), summarize the results to the user.