Review.prompt
---
description: "3-pass procedural code review: correctness, security, quality"
application: "When reviewing code changes, pull requests, or auditing specific files"
mode: agent
agent: Alex
model: claude-opus-4-6
currency: 2026-04-21
---
# /review - Code Review
3-pass review with structured findings. Create a TODO list for all passes. Mark each in-progress before starting, completed immediately after finishing.
## Step 1: Scope
1. Identify the files to review (from user input, diff, or changed files)
2. Read each file fully before reviewing
## Step 2: Pass 1 — Correctness
For each file:
1. Check logic errors, off-by-one, null/undefined paths, race conditions
2. Check error handling: are failures caught and reported?
3. Check edge cases: empty inputs, boundary values, unexpected types
4. Record findings with file, line, severity (critical/warning/info)
## Step 3: Pass 2 — Security
For each file:
1. Check OWASP Top 10: injection, broken auth, sensitive data exposure
2. Check input validation at system boundaries
3. Check credential handling: no secrets in code, no log leakage
4. Record findings with file, line, severity
## Step 4: Pass 3 — Quality
For each file:
1. Check naming clarity, function length, DRY violations
2. Check test coverage for changed logic
3. Check consistency with existing codebase patterns
4. Record findings with file, line, severity
## Summary
After all passes, generate:
- Files reviewed (count)
- Findings by severity (critical / warning / info)
- Findings by pass (correctness / security / quality)
- Top 3 issues requiring immediate attention
- Overall assessment (approve / request changes)when to use it
Community prompt sourced from the open-source GitHub repo fabioc-aloha/alex-cognitive-architecture (NOASSERTION). A "Review.prompt" style prompt — adapt the placeholders and specifics to your task. Imported as-is and not independently retested here, so check the output before relying on it.
tags
codingcommunitydeveloper
source
fabioc-aloha/alex-cognitive-architecture · NOASSERTION
more in Coding
Coding✓ tested
Senior code review (strict mode)
senior staff engineer running a merciless but fair review
Coding✓ tested
Debug by hypothesis, not by guessing
debugging partner who forms theories before touching code
Coding✓ tested
Generate tests from described behavior
test engineer who writes tests that would actually catch regressions