agent-qa-debug-fix

Agent QA Debug Fix

Overview

Repair a failed Agent QA run from recorded evidence and the relevant local source. Treat the classifier as a hypothesis, make the smallest justified change, and verify the narrowest affected behavior without rewriting a test merely to conceal a real defect.

When to Use

• A failed Agent QA run has already been triaged and now requires a code or YAML repair.
• Artifacts and logs point to a test, hook, product, runtime, or agent-behavior defect.
• A proposed fix must be verified with the narrowest Agent QA or unit-test rerun.
• The user asks to self-heal or update a stale Agent QA definition from evidence.

Preconditions and Approval Boundary

• Confirm the repository, workspace, target environment, and files the user authorizes you to modify.
• Inspect the planned test's external side effects before rerunning it; obtain explicit confirmation for production-facing, destructive, or irreversible actions.
• Preserve unrelated user changes and keep the patch limited to the evidenced failure.
• Do not expose credentials or sensitive application data from artifacts and logs.

Workflow

1. Start with evidence collection:
   • agent_qa_get_run
   • agent_qa_get_run_steps
   • agent_qa_get_run_artifact
   • agent_qa_get_run_logs
   • agent_qa_get_run_execution_logs
2. Call agent_qa_classify_failure and treat its category as a hypothesis, not a verdict.
3. Identify the failing surface: test definition, hook, application under test, runtime infrastructure, or agent behavior.
4. Inspect the relevant local files directly. Do not infer patches from artifacts alone.
5. Explain the evidence-to-change link, then apply the smallest code or YAML change that accounts for the evidence.
6. Validate any changed Agent QA definition before execution.
7. Re-run the narrowest affected Agent QA test, suite, hook, or unit test within the approved environment.
8. Report the root cause, changed files, verification command or MCP action, result, and remaining risk.