janekbaraniewski/openusage/review-design
Claude Code and agent skill by janekbaraniewski/openusage
- community
Checking that the SKILL.md file is still in the repository.
Review a design doc against the actual codebase, find inconsistencies, and quiz the user on needed fixes. Loops until all issues are resolved. Use after /design-feature and before /implement-feature.
Install
npx vskill@latest install janekbaraniewski/openusage/review-designSKILL.md
OpenUsage Design Doc Reviewer
Invocation: When a user wants to validate a design doc against the current codebase before implementing it.
Input: Design doc path or feature name (resolved to docs/<NAME>_DESIGN.md).
Phase 0 — Load Design
- Find the design doc. If path not given, search
docs/*_DESIGN.md. - Read the full design doc. Extract:
- Problem statement
- Affected subsystems
- Type definitions and interface changes
- Implementation tasks with file lists
- Confirm with the user which design doc to review.
Phase 1 — Codebase Audit
For each subsystem listed in the design doc's impact analysis:
- Read the primary files from
docs/skills/design-feature/references/subsystem-map.md. - Read every file referenced in the implementation tasks.
- Build a discrepancy list by checking each item in
references/review-checklist.md.
Focus on concrete mismatches. Don't flag style preferences or hypotheticals.
Phase 2 — Quiz Loop (CORE)
Present findings one at a time as questions. For each discrepancy:
Rules for the quiz loop:
- Present 1-3 issues per round (group related issues).
- Wait for the user's answer before proceeding.
- After the user answers, apply their decision:
- If (a): edit the design doc immediately.
- If (b): note it as "intentional change" — no edit needed.
- If (c): discuss, then apply the agreed resolution.
- After resolving a batch, re-scan affected areas for new issues introduced by the edits.
- Repeat until no discrepancies remain.
Phase 3 — Final Verification
When no more issues are found:
- Re-read the design doc (it may have been edited during the quiz).
- Verify all implementation tasks still reference valid files and types.
- Confirm the task dependency order is still correct.
- Report:
Rules
- Never skip Phase 1. Always audit the codebase — don't trust the design doc blindly.
- Never auto-fix without asking. Every discrepancy goes through the quiz.
- Always cite file:line. Vague references like "in the types file" are not acceptable.
- Stop if the design doc is fundamentally broken (e.g., references a subsystem that doesn't exist). Tell the user to re-run
/design-feature. - Don't review code style. Focus on structural correctness: types, interfaces, function signatures, file locations, config schema.
- Group related issues. If three fields on the same struct are wrong, that's one issue, not three.
Security
- Security scanPassed, 95 of 100Pattern checks on the skill's files.
- Model reviewPassed
Quality evaluation
- Task evaluationNot runNo evaluation yet. Run one in Skill Studio.
Source
- SKILL.md in the repositoryCheckingdocs/skills/review-design/SKILL.md
Popularity
- 7-day trend score
- 0
Refreshed 19 Mar 2026, 7 months ago.
Works with
Any agent vskill installs to: 56 targets in all. These 39 have a page on this site.
- Claude Code
- Cursor
- GitHub Copilot
- Windsurf
- Codex
- Gemini CLI
- Cline
- Amp
- Roo Code
- Goose
- Continue
- Kilo Code
- Junie
- Trae
- OpenHands
- Qwen Code
- AI Agent
- Tabnine Chat
- Sourcegraph Cody
- Amazon Q
- Firebase Genkit
- Sweep
- Devin
- Phind
- Mentat
- Claude Desktop
- Aider
- GPT Pilot
- Plandex
- bolt.new
- v0
- Replit Agent
- Anse
- TypingMind
- BetterChatGPT
- LibreChat
- Warp AI
- Pieces
- Fig AI
Embed badge
[](https://verified-skill.com/skills/janekbaraniewski/openusage/review-design)