---

name: debug-triage description: "Systematically reproduce, diagnose, and fix bugs; use when debugging failures, crashes, or test errors (trigger keywords: debug, bug, failure, crash, flaky test)."

Debug Triage

What this skill does

• Reproduces issues, isolates root causes, and applies minimal fixes.
• Verifies fixes with targeted tests and documents findings.
• Produces a concise bug report when reproduction is possible.

When to use it

• Trigger phrases: "debug", "bug", "failure", "crash", "flaky test".
• Use for runtime errors, test failures, or unexpected behavior.

Step-by-step workflow

1. Gather context: expected vs actual behavior, error output, logs, environment details, and recent changes.
2. Reproduce with the smallest reliable steps; if not reproducible, ask for missing info or add targeted logging.
3. Trace the code path and form specific root-cause hypotheses (include config, data, and concurrency where relevant).
4. Validate hypotheses with evidence from code, logs, or targeted tests; rule out alternatives.
5. Apply a minimal fix aligned with project patterns; avoid unrelated refactors.
6. Verify: rerun reproduction steps and relevant tests; note any regression risks.
7. Document root cause, fix, tests, and remaining risks or follow-ups.

Expected outputs / formatting

• Repro steps with expected vs actual behavior and observed output.
• Environment details (OS/version/config) when relevant.
• Root cause explanation with file references.
• Hypotheses tested (brief) and evidence used.
• Fix summary and tests run.
• Remaining risks or follow-ups.

Example prompts

• "Debug this crash and propose a fix."
• "Investigate why this test is failing."
• "Triage this flaky test and suggest a minimal change."