Engineering

Code Review Assistant

Turn AI into a structured pull request reviewer that catches risky changes, flags maintainability issues, and suggests missing test coverage.

Open in System Prompt Generator

Overview

A structured system prompt that configures an AI assistant to perform disciplined code reviews. Instead of vague feedback, the assistant follows a consistent review framework: assessing correctness, security implications, test coverage, and maintainability — then delivering findings in a format your team can act on immediately.

How to use this resource

Copy the diff or file

Paste the raw diff from your PR, or the full file if reviewing a new addition.
Run through the assistant

Send the pasted code to your AI model with this prompt as the system instruction.
Triage action items

Work through the numbered action items. Dismiss or address each one before merging.
Iterate on re-diffs

After addressing feedback, re-run the revised diff to confirm all action items are resolved.

Why This Works

Structured headings force the AI to address each review dimension rather than producing a general impression
Separating correctness, security, and maintainability prevents one dimension from dominating the output
Numbered action items make findings directly assignable — vague feedback creates follow-up questions
Instructing the model not to summarize what the code does eliminates the most common AI review waste

Best for

Teams without dedicated code review tooling
Solo developers who want a disciplined second opinion
Pull requests touching security-sensitive paths
Onboarding reviews where thoroughness matters

Not for

Style-only linting — use a formatter or linter instead
Auto-fixing issues — review output is read-only by design
Performance profiling without benchmark data

Use cases

Reviewing pull requests before merging to main
Catching security regressions in authentication or authorization code
Ensuring new engineers receive consistent, structured feedback
Auditing legacy code before a major refactor

FAQ

Does this code review system prompt catch every security issue in a PR?

It structures the review, it doesn't guarantee coverage. The Security heading prompts for "injection risks, exposed credentials, unsafe deserialization, or permission issues," but an AI reviewer can miss things, and the notFor line keeps output "read-only by design." It's a disciplined second opinion you run in ChatGPT or Claude; a human still owns the merge and any real security sign-off.

How do I act on the output from this structured review prompt?

Work the Action Items — the last heading is a numbered list of specific changes to make before merging, which the whyThisWorks notes are "directly assignable." Triage each, address it, then re-run the revised diff to confirm resolution. Empty sections read "No issues found." so nothing is silently skipped. The prompt organizes feedback; you and your team apply the fixes.

Why does this reviewer skip describing what the code does?

By design — the closing rule is "Do not summarize what the code does — the author already knows," and the whyThisWorks calls that the most common AI review waste. Every heading is a findings dimension instead: Correctness, Security, Test Coverage, Maintainability. The system prompt front-loads that framework so the pass spends tokens on issues, not narration you'd skim past.

Customize This Resource

Opens this resource in System Prompt Generator with the fields pre-filled. Customize role, tone, and behavior rules — then generate your prompt.

Open in System Prompt Generator

Prompt Template

Copy it as-is, or use Open in System Prompt Generator to load it pre-filled and customize it with your own context.

You are a senior software engineer performing a structured code review.

For every file or diff you review, provide your findings under the following headings:

## Summary
A one-sentence assessment of the change.

## Correctness
Identify logic errors, off-by-one issues, or incorrect assumptions.

## Security
Flag any injection risks, exposed credentials, unsafe deserialization, or permission issues.

## Test Coverage
Note missing unit tests, edge cases not covered, or branches that need regression coverage.

## Maintainability
Comment on naming clarity, function length, abstraction quality, and technical debt introduced.

## Action Items
A numbered list of specific changes the author should make before merging.

Keep each section concise. If a section has no findings, write "No issues found." Do not summarize what the code does — the author already knows.

More resources from System Prompt Generator

Resource

Bug Triage Assistant

Convert scattered bug notes, Slack messages, or user complaints into structured engineering tasks with reproduction steps, severity, and root cause hypothesis.

Engineering

Resource

Customer Support Agent

Configure AI to answer support questions within your actual policy boundaries — not generic best-guess answers.

Support

Resource

Research Synthesis Assistant

Extract and compare findings from multiple sources without collapsing them into a single blended perspective.

Research

Resources that pair well

Resource

Product Manager Role Prompt

A complete 'act as a product manager' role prompt — the perspective, responsibilities, and decision criteria a real PM brings, not just the job title.

Product

Resource

Agent Task Template

A reusable AI agent task template with variables for objective, context, available tools, constraints, success criteria, failure handling, and output format.

AI Agents

Resource

Agent Instruction Prompt Formatter

Formats fuzzy agent instructions into a structured prompt with objective, available tools, constraints, success criteria, and failure handling.

AI Agents

Related tools

Tool

System Prompt Generator

Build structured system prompts from role, tone, constraints, and model target.

Prompt Builders

Tip: Save time by exploring related resources and tools that integrate with this resource.