Code review: aim for coverage, don't filter for me
A ready-to-use “Programming” AI prompt template. Replace the (bracketed) parts below with your own needs.
Prompt template
Review the code below. This round's goal is **coverage**, not brevity. [Code] (paste the code you want reviewed) [Background] - Language and framework: (e.g., TypeScript + React) - What this code is for: - Runtime environment and scale: (e.g., handles 100,000 requests a day) - The part I'm most worried about: (may leave blank) [Review requirements] Report **every** issue you find, including ones you're unsure about or think are low severity. Don't filter by importance at this stage — I'll rank them myself later. Better to report one extra that I later dismiss than to miss a real bug. Use this format for each finding: - **Location**: line number or function name - **Issue**: concretely what will go wrong - **Trigger scenario**: what input or state causes it - **Severity**: high/medium/low - **Confidence**: high/medium/low - **Suggested fix**: concrete enough to apply directly [Specially check these categories] 1. Edge conditions (null, zero, huge values, empty arrays) 2. Concurrency and race conditions 3. Error handling and exception paths 4. Resource leaks (connections, files, memory) 5. Security (injection, permissions, sensitive data leakage) Finally give me a summary table sorted by "severity x confidence." Use Traditional Chinese with Taiwan usage.
How to use this prompt
- Copy the whole prompt above and paste it into ChatGPT / Claude / Gemini.
- Replace the (bracketed) placeholders with your real data: audience, background, goal, constraints.
- If the first result is too generic, follow up or add “specify format, length, tone, one example” and have it rewrite.