원클릭으로
sspec-review
User acceptance and feedback loop. Handle argue-improve cycles until user is satisfied.
Codex 또는 Claude로 설치 이 Prompt를 복사해 Codex, Claude 또는 다른 어시스턴트에 붙여 넣으면 Skill 페이지를 검토하고 설치를 진행할 수 있습니다.
메뉴
User acceptance and feedback loop. Handle argue-improve cycles until user is satisfied.
Codex 또는 Claude로 설치 이 Prompt를 복사해 Codex, Claude 또는 다른 어시스턴트에 붙여 넣으면 Skill 페이지를 검토하고 설치를 진행할 수 있습니다.
SOC 직업 분류 기준
Pre-scan Markdown files before reading. Use `sspec tool mdtoc` to get file size + heading structure with line numbers (L<n> format) before reading long docs, SKILLs, or spec-docs. Eliminates blind reads.
Assess scale, create change, fill spec.md + design.md, align with user. Use after clarify when ready to define the solution.
Execute tasks from tasks.md. Implement code changes, update progress. Use after plan approval.
Break design into concrete tasks. Fill tasks.md with file-level execution plan. Use after design alignment.
Build shared understanding through dialogue and investigation. Produces Problem Statement + direction sketch. Reusable posture, not rigid phase.
Structured, efficient agent-user synchronization at decision points. Formalized exchange over prose.
| name | sspec-review |
| description | User acceptance and feedback loop. Handle argue-improve cycles until user is satisfied. |
| metadata | {"author":"frostime","version":"4.0.0"} |
Collect user feedback, iterate improvements, close the loop.
User feedback ─→ Classify (@feedback-class) ─→ Act ─→ @align "Fixed. Check again?"
↑ │
└────────────────────┘
(repeat until user satisfied)
Before touching any file, output the classification:
@feedback-class: <minor-fix | amend | follow-up | supersede>
Reason: <one sentence>
The single test to apply:
Can the original spec/design still accurately predict the post-change code? YES → minor-fix | NO → amend → revision required
| ✅ minor-fix | ❌ amend (revision required) |
|---|---|
| 变量 / 函数命名调整 | 新增验收条件 / Behavior Contract |
| typo / 文案修正 | 新增验证 / User Check / 日志 / 错误分支 |
| 明显 bug(无行为变更) | 新增用户可见行为 |
| 已有验收边界内的边界修复 | 修改范围边界 |
| 任何让原 spec 无法完整预测最终代码的反馈 |
| Class | Action |
|---|---|
| minor-fix | Fix directly or add Feedback Tasks if non-trivial |
| amend | See amend protocol below |
| follow-up | @align user before opening a new change with prev-change reference |
| supersede | @align user before marking current change BLOCKED and opening a replacement |
📚 sspec howto handle-review-scope-change
sspec change scaffold revision <change> --title "..." — create revision file firstspec.md frontmatter reference::
- source: ".sspec/changes/<change>/revisions/NNN-xxx.md"
type: "revision"
note: "<one-line summary>"
Feedback Tasks block in tasks.md with header linking the revision:
### Feedback Tasks (→ [NNN-xxx](./revisions/NNN-xxx.md))
- [ ] ...
Use Feedback Tasks only for work that still belongs to the current change. Do not use it as a dumping ground for follow-up or replacement work.
When memory.md has a Git Baseline section, use it as the review anchor.
📚 sspec howto review-git-baseline
For independent review, consider subagent audits to avoid context blind spots.
📚 sspec howto make-subagent-audit | Directive: @subagent-audits
When user is satisfied:
[x]; update status: REVIEW → DONE@align user about creating/updating a spec-doc@memoryDOING (implement complete) → REVIEW (user reviewing)
→ DONE (user satisfied)
→ DOING (needs more work → implement feedback tasks → REVIEW again)
→ PLANNING (accepted redesign → back to design/plan)
→ BLOCKED (user chooses replacement or external blocker)
FORBIDDEN: PLANNING→DONE, DOING→DONE — never skip REVIEW.