Code review with explicit constraints
Keeps AI code review scoped to your constraints so it does not invent policy or fake severity scores.
Review the following change under explicit constraints. Do not invent org policy.
Constraints:
- Languages/frameworks: [list]
- Must not: [e.g. add new dependencies, break public API]
- Security focus: [auth, secrets, injection, none]
- Performance focus: [hot path, allocations, none]
- Style: [link or “match existing file”]
Diff or files:
```
[paste]
```
Rules:
- Findings first; no overall “quality score” out of 100.
- Mark confidence: high / medium / low.
- If you lack context, ask a question instead of guessing.
Return:
1) Summary of the change (3 sentences)
2) Findings table (severity: blocker/major/minor/note | location | issue | suggestion)
3) Questions for the author
4) Suggested test planAbout this template
Reason
Keeps AI code review scoped to your constraints so it does not invent policy or fake severity scores.
Outcome
A review checklist and comment draft tied to stated rules.
Insight
Ask for severity as labels you define—not a pseudo-numeric quality score for the whole codebase.
These templates are starting points for your own work. They are not ranked, scored, or sold. AIScoreboard does not crown a “best prompt.” Money never buys rank, and these pages have no Stripe checkout or paywall.