test-result-analyzerlisted

# Test Result Analyzer An L2 Truth-Execution skill. Where `regression-test-runner` tells you **whether** tests failed and `cross-run-consistency` tells you **whether they agree with themselves**, this skill tells you **why** each failure happened and **what to do about it**. It turns a pile of red into a classified pile, and the classification is what downstream tools (`release-decision-engine`, `learning-loop-engine`) use to decide if a failure is a ship- blocker or an operational nuisance. The most common failure mode of a result analyzer is silently dropping failures into a "we'll figure it out later" bucket. This skill's gate contract is specifically designed to prevent that — **no failure leaves the classification step without a label**, and every label with low confidence is surfaced for human triage rather than auto-ticketed as a bug. ## When You're Invoked - **PIPELINE-5 step 4** — after `regression-test-runner` and `uat-executor` have produced their raw outputs, before `release-decision-engine` computes its verdict. The decision engine reads the classified failure set from this skill's output, not the raw reports. - **PIPELINE-6 step 3** — same position in the release-track pipeline, after the full regression + chaos runs. - **On demand** as `/vibeflow:test-result-analyzer <report-path-or-glob>`. ## Input Contract | Input | Required | Notes | |-------|----------|-------| | Primary report | yes | Exactly ONE of: `uat-raw-report.md` / `regression-repo