regression-test-runnerlisted

# Regression Test Runner An L2 Truth-Execution skill. Its job is to tell the delta since the last green run: which tests regressed, which got fixed, which are just flaky, which are still stuck red. The baseline is the memory that makes that delta meaningful, and the gate contract is the rule that keeps the baseline honest — **the baseline is never promoted forward over a failing P0 test**. Where `e2e-test-writer` generates tests and `uat-executor` runs scenarios, this skill exercises the committed test suite. It doesn't write code, it doesn't touch production, it doesn't try to interpret a failure. That's what `test-result-analyzer` is for (Sprint 3 downstream). ## When You're Invoked - **PIPELINE-2 step 4** — after a code change, before merge. Default scope is `smoke`. Fast feedback loop. - **PIPELINE-5 step 2** — before a release tag. Default scope is `full`. Full audit. - **On demand** as `/vibeflow:regression-test-runner [--scope <s>] [--since <sha>]`. - **From `release-decision-engine`** when the decision engine needs a fresh baseline snapshot for the GO/CONDITIONAL call. ## Input Contract | Input | Required | Notes | |-------|----------|-------| | Trigger metadata | yes | One of `pr / push / release / manual` — determines default scope | | `regression-baseline.json` | optional but preferred | Previous baseline. Absent → cold start (see §6). Present but stale → staleness block (see `references/baseline-policy.md`). | | `--scope` | optional | `smoke / full / i