Windsurf rules

1---
2name: gate-trace
3model: haiku
4description: "Deterministic traceability quality gate — reads coverage matrix + blind-spot data, applies decision rules with oracle confidence downgrade, emits PASS/CONCERNS/FAIL/WAIVED verdict. Use before release-branch to gate merges on traceability."
5---
6 
7# Gate Trace
8 
9Deterministic quality gate that combines traceability coverage and blind-spot data into a single PASS/FAIL/CONCERNS/WAIVED decision before release.
10 
11## Decision Rules
12 
13| Rule | Condition | Verdict |
14|------|-----------|---------|
15| R1 | Any undone story with 0 code tags | FAIL |
16| R2 | Any story done but no verify evidence | CONCERNS |
17| R3 | P0 story (top WSJF quartile) with 0% coverage | FAIL |
18| R4 | Overall coverage < 60% | CONCERNS |
19| R6 | P0 story + eNN-TEST_PLAN_LATEST.md exists + zero SC-eNNsYY-P0-* in test files | CONCERNS |
20| R5 | Overall ≥ 80% + no critical gaps + all verify → PASS | PASS |
21 
22## Oracle Confidence Downgrade
23 
24TEA-inspired: if trace links rely on heuristics rather than explicit tags, confidence drops.
25 
26| Heuristic Ratio | Downgrade |
27|-----------------|-----------|
28| > 50% links from heuristics (file-name or task-reference) | One level (PASS→CONCERNS, CONCERNS→FAIL) |
29| > 80% links from heuristics | Two levels (PASS→FAIL, CONCERNS→FAIL) |
30 
31## Process
32 
331. **Pre-flight:** Ensure `scripts/trace-stories.sh --json` has been run (produces `specs/traceability-matrix.json`). If not present, run it.
342. Ensure `scripts/check-blind-spots.sh` has been run (produces `specs/blind-spots.json`). If not present, run it.
353. Read `specs/traceability-matrix.json` and `specs/blind-spots.json`.
364. Apply decision rules R1–R5 in order (first match wins).
375. Apply oracle confidence downgrade based on the heuristic link ratio from the matrix's `oracle_stats`.
386. **Drift check (e39s03):** If `specs/drift-report.json` exists and has suspect links, mark verdict as CONCERNS with note: "Drift detected — some implementing files are newer than their specs. Run scripts/check-spec-drift.sh for details."
397. Output verdict + rationale to stdout.
408. **Adversarial refute check (e45s32):** Before emitting PASS, attempt to **refute** the verdict — list at least one concrete traceability gap that would block merge if it were real. If the gap is real, downgrade the verdict. Rubber-stamping is prohibited; every PASS must survive one refutation attempt.
419. **Completeness critic (e45s05)** — Run adversarial gap-finding:
42 
43```bash
44bash scripts/lib/completeness-critic.sh
45```
46 
47Classify output as **BLOCKER / WARNING / FILLED**. **BLOCKER overrides any PASS verdict → FAIL.** Append critic summary to rationale.
4810. Update `specs/execution-status.yaml` with gate-trace result.
49 
50## Verdict Semantics
51 
52| Verdict | Meaning | Action |
53|---------|---------|--------|
54| PASS | All gates satisfied | Proceed with merge |
55| CONCERNS | Non-critical issues found | Requires explicit human override in `state.yaml` |
56| FAIL | Critical traceability gap | Block merge — fix gap first |
57| WAIVED | Cannot evaluate (missing inputs) | Skip gate — data not available |
58 
59## Output Format
60 
61```yaml
62gate_trace:
63  verdict: PASS|CONCERNS|FAIL|WAIVED
64  generated_at: "<ISO 8601>"
65  rationale: "<human-readable explanation>"
66  heuristic_ratio: <0.0–1.0>
67  downgrade_applied: <true|false>
68```
69 
70## Integration Points
71 
72- **release-branch SKILL.md** — pre-PR gate (FAIL blocks merge).
73- **sync-skills.yml** — runs `trace-stories.sh --strict` as its "Traceability gate" step (FAIL
74  blocks pipeline), which shares this skill's coverage data (`traceability-matrix.json`) but
75  does not invoke a gate-trace PASS/CONCERNS/FAIL/WAIVED verdict directly.
76- **verify-work SKILL.md** — runs blind-spot checks; gate-trace consumes the result.
77 
78## References
79 
80- BMAD TEA traceability approach (market survey 2026-07-02).
81- scripts/trace-stories.sh (e38s01) — produces traceability-matrix.json.
82- scripts/check-blind-spots.sh (e38s04) — produces blind-spots.json.
83- specs/execution-status.yaml — output target for gate result.
84 
85## Handoff
86 
87Gate: READY → next: release-branch (final step before merge)
88Writes: state.yaml handoff.next_skill = release-branch
89 
90## Verify
91 
92→ verify: `test -x scripts/run-gate-trace-verify.sh && bash scripts/run-gate-trace-verify.sh --self-test && echo OK`
93