modrexio/modrex / .agents/skills/ai-review
ai-review
VerifiedReview a diff for AI-shaped code before a PR, commit, or final summary
ai-review
Use this skill before proposing a PR, commit, or final response for a non-trivial code change.
Start with blocker-level issues from .agents/rules/ai-danger-patterns.md.
Review target
Default to git diff HEAD.
If the user provides a path or range, review only that target.
Blocker checklist
Check first for the five dangerous AI patterns:
- Silent catch / silent fallback
- Generic helpers replacing game-specific logic
- Defensive checks that hide broken invariants
- Deep
if/elsenesting - Fake abstractions / speculative future-proofing
Cleanup checklist
Then check for:
- obvious WHAT-comments
- useful structural/header comments wrongly removed
anycasts- unnecessary
useMemooruseCallback - derived data stored in React state
- unrelated formatting
- hardcoded renderer colors (use semantic tokens from
index.css) - direct API calls that bypass project caches (
modCache.ts,thumbnailCache.ts) - bare JSX strings (use
t('key')) - missing Tauri command wiring (all three places)
- user-facing behavior missing from CHANGELOG when appropriate
- code that contradicts the nearest
AGENTS.mdabove it
Severity
Use three levels:
- Blocker: dangerous AI pattern, likely bug, broken invariant, wrong project pattern, unsafe filesystem/archive behavior.
- Cleanup: maintainability issue, obvious AI-shaped code, useless comment, unnecessary abstraction.
- Note: acceptable but worth mentioning.
Output
Return:
- Blockers
- Cleanups
- Notes
- Suggested next command or skill
If there are no issues:
Diff is clean against the ai-review checklist.
Do not invent problems. Cite the exact file and function when possible.
Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| receiving-code-reviewobra/superpowersUse when receiving code review feedback, before implementing suggestions, especially if feedback seems unclear or technically questionable - requires technical rigor and verification, not performative agreement or blind implementation | Stars: 321k | Updated | Claude CodeCodex | Verified | Install |
| requesting-code-reviewobra/superpowersUse when completing tasks, implementing major features, or before merging to verify work meets requirements | Stars: 321k | Updated | Claude CodeCodex | Verified | Install |
| using-git-worktreesobra/superpowersUse when starting feature work that needs isolation from current workspace or before executing implementation plans - ensures an isolated workspace exists via native tools or git worktree fallback | Stars: 321k | Updated | Claude CodeCodex | Verified | Install |
| verification-before-completionobra/superpowersUse when about to claim work is complete, fixed, or passing, before committing or creating PRs - requires running verification commands and confirming output before making any success claims; evidence before assertions always | Stars: 321k | Updated | Claude CodeCodex | Verified | Install |