bug-report-writing
Use when any case fails or you reject a criterion - what goes in the review_criterion note, the failed case and the numbered need_revision comment, with severity
- License 许可
- Apache-2.0
- 最近更新
- 2026-11-15
- Stars
- ★ 122
- Forks
- 18
01 前言 Front Matter
- name
- bug-report-writing
- category
- qa
- description
- Use when any case fails or you reject a criterion - what goes in the review_criterion note, the failed case and the numbered need_revision comment, with severity
03 SKILL.md 正文
Bug Report Writing
A defect's evidence lives in three places, each with its own job — never duplicate the full report into all three:
| Place | Contents |
|---|---|
| Failed case | actual = the observed output, verbatim. evidence = exact command/flow plus frequency, e.g. "3/3". |
| Rejected criterion note | Expected (quote the criterion), actual, reproduction, environment: short SHA, URL/base, width or device, browser/runtime. |
| need_revision comment | A numbered index, one line per defect: 1. [AC2 · case "export rejects >10k rows"] 500 instead of 422 — S2. Keep it inside the 15-line comment budget (concise-board-comments) — the reproduction lives in the criterion note and case evidence, not here. |
What fails the task
A task fails when:
- (a) a criterion or a human requirement comment is unmet;
- (b) the team UI floor is broken, by a measured threshold: horizontal overflow at any of 360/768/1024/1440, a touch target <44px at 360, an input font <16px at 360, body contrast <4.5:1, a missing loading/empty/error state on a changed view, an uncaught console error or failed request in the flow, a broken image/icon, a critical/serious axe violation inside a changed screen;
- (c) a crash, data loss, a security or authz hole, or a regression of adjacent behaviour;
- (d) a Blocker or High deviation from the approved design the task builds (design-conformance) — Medium and Nitpick are notes.
Anything else — taste, polish beyond the floor — is a note on the criterion, never a failure.
Rules
- Title names the problem, not the fix, under 60 characters: "Export of >10k rows returns 500", not "Fix export pagination".
- Isolate to the first failing step; give the minimal reproduction, not the whole scenario.
- Separate observation from speculation. A location from a stack trace or a
grep_codehit is fine (the named debugging exception); a proposed fix is not your job. - Severity, worst wins: S1 data loss/security hole > S2 broken flow (criterion unmet) > S3 degraded UX (UI-floor violation) > S4 cosmetic (never fails the task on its own — see qa-criterion-verdicts).
- Never invent a file path. A screenshot has none; describe width/URL/what-it-showed instead (ui-visual-evidence).
- One defect per case/criterion entry — do not bundle unrelated failures into one note.
Worked Example
❌ "Export broken" — no repro, no expected, no severity.
✅
Case: export-pagination-10k (failed, 3/3)
actual: HTTP 500, body {"error":"internal"}
evidence: curl -X GET "$BASE/api/export?limit=10001" -H "Authorization: Bearer $TOKEN" — ran 3x, same result
Criterion note (AC2, rejected):
Expected: "requests over the 10,000-row limit return 422 with a clear message" (AC2)
Actual: 500 with no message, see case export-pagination-10k
Repro: curl -X GET "$BASE/api/export?limit=10001" ...
Env: SHA a1b2c3d, http://localhost:5173, desktop 1440
need_revision comment:
1. [AC2 · case "export-pagination-10k"] 500 instead of 422 at limit+1 — S2
04 同仓库更多 Skills
visual-craft
暂无简介
ux-copy
Use when you write any words a user reads in a mockup or a hand-off - content language and voice, buttons, labels, errors, empty states, confirmations, numbers and dates, and the longest-string check
screen-mockup-html
Use when you draw screens as the design task's HTML mockup - ONE canvas document holding every screen and every variant side by side, the sanitizer-safe recipe, phone and wide frames, the states grid, static control states, and the loopback self-check with screenshots at 375, 768 and 1440 attached to the task
performance-awareness
How the performance score moves for your role. Use when the score context shows a declining trend, or before handing off work you have not self-checked against every criterion.
design-variants
Use when a screen design needs variants - how many, variants as side-by-side rows on the one canvas, what makes two variants genuinely different, recommending one, how the human's choice is read, and narrowing to the chosen one in revision
design-system-authoring
Use when you derive, change or extend a design system - where tokens live per stack, the DTCG token tree, role-based names for both themes, base vs layer vs drift, the DESIGN.md section order, the inventory format, the contrast check and the lint every proposal returns