test-priority-enginelisted

# Test Priority Engine An L2 Truth-Execution skill. It doesn't run tests — it decides what order `regression-test-runner` (or a human operator) should run them in, and it caps the plan to a time budget that matches the CI stage the user is in. Fast feedback is a product; exhausting the test suite every commit isn't. The risk model is deterministic and documented in `references/risk-model.md`. The mode budgets are in `references/mode-budgets.md`. Both files are load-bearing — a weak priority is how "the tests I should have run" becomes "the bug I shipped". ## When You're Invoked - **PIPELINE-2 step 1** — before `regression-test-runner` on an incremental PR run. The plan produced here becomes the input `--scope=incremental` for the runner. - **PIPELINE-5 step 1** — before the pre-release regression run. The plan is used as a scheduling hint so the highest-risk tests run earliest, and a failure trips the gate before the long tail. - **On demand** as `/vibeflow:test-priority-engine [--mode <m>] [--since <sha>]`. - **From `regression-test-runner`** when it needs the ordering for its own Step 1 scope resolution. ## Input Contract | Input | Required | Notes | |-------|----------|-------| | Changed files list | yes | From `git diff --name-only <since>..HEAD` or an explicit file list. Empty list → "no risk signal"; the skill emits a full-suite plan sorted by priority-only fallback and WARNs. | | `regression-baseline.json` | optional but preferred | Used for baseline fa