How do I install the AI Review skill?

Run npx skillmds@latest add 89jobrien/ai-review in your terminal (requires Node.js), paste this page's agent-chat prompt into Claude, Cursor, or any MCP-connected agent, or download the SKILL.md file and copy it into your agent's skills directory.

Is AI Review safe to use?

This skill has not completed SkillMD's automated safety review yet. SkillMD never runs a skill's scripts for you; review the SKILL.md before installing.

Which AI agents work with AI Review?

This skill is tagged as working with Claude Code, Claude.ai, OpenAI Codex. SKILL.md is an open format, so most agents that read a skills directory can load it too.

Is AI Review free to use?

Yes. Installing skills from SkillMD is free, and the skill stays under its author's original license.

Who published AI Review?

89jobrien (@89jobrien) published this skill. Their other Agent Skills are listed on their SkillMD profile.

AI Review by 89jobrien — Agent Skill

Name: AI Review
Author: 89jobrien

AI Review

This skill should be used when the user asks to "review this diff", "ai review", "security review my changes", "review before push", "check my code for issues", or wants an LLM-based security and correctness review of uncommitted or unpushed changes.

89jobrien Updated Sep 17, 2026

File contents

ai-review

Run an LLM-based security and correctness review of the current diff vs a base branch. Surfaces findings grouped by severity. Writes a markdown report to .ai-logs/.

Workflow

1. Get the diff

Default base branch is main. Accept an explicit base if provided.

git diff main...HEAD

If HEAD is already on main or there is no diff, fall back to staged changes:

git diff --staged

If both are empty: inform the user there is nothing to review and stop.

2. Get the current short SHA

git rev-parse --short HEAD

Use this as <sha> in the report filename.

3. Send to LLM for review

Compose a prompt with the following structure:

You are a senior engineer performing a security and correctness review of a code diff.

Review the diff below and produce a structured report with findings grouped by severity:
- CRITICAL: security vulnerabilities, data loss, auth bypasses, path traversal, injection
- HIGH: correctness bugs, panics in production paths, missing error handling, race conditions
- MEDIUM: performance issues, missing cleanup, unclear invariants, incomplete implementations
- LOW: style issues, dead code, missing docs, minor inefficiencies

For each finding:
- Severity: CRITICAL | HIGH | MEDIUM | LOW
- File + line reference (if determinable from diff)
- Finding: one-sentence description
- Detail: 2-3 sentences of explanation
- Recommendation: concrete fix

If no findings at a severity level, omit that section.
End with a one-paragraph summary verdict.

DIFF:
<diff content>

4. Write report

Ensure .ai-logs/ directory exists, then write the report:

.ai-logs/ai-review-<sha>.md

Report format:

# AI Review — <sha>

**Base:** main  **Date:** <date>

## CRITICAL

### <Finding title>
**File:** `path/to/file.rs:42`
**Finding:** ...
**Detail:** ...
**Recommendation:** ...

## HIGH
...

## Summary

<verdict paragraph>

If the directory does not exist, create it before writing.

5. Surface results

Print the findings summary to the terminal grouped by severity with counts:

SEVERITY   COUNT
CRITICAL   0
HIGH       2
MEDIUM     3
LOW        1

Report: .ai-logs/ai-review-<sha>.md

If CRITICAL or HIGH findings exist, highlight them inline before the summary table.

Base Branch Override

If the user specifies a different base (e.g. "review against next"), use that ref instead of main for the diff command.

Pairing with cargo-gate

Run cargo-gate before ai-review on Rust projects to ensure the diff is clean first. The review is most useful on a diff that already compiles and passes clippy.

---name: ai-reviewdescription: This skill should be used when the user asks to "review this diff", "ai review", "security review my changes", "review before push", "check my code for issues", or wants an LLM-based security and correctness review of uncommitted or unpushed changes.---# ai-reviewRun an LLM-based security and correctness review of the current diff vs a base branch.Surfaces findings grouped by severity. Writes a markdown report to `.ai-logs/`.## Workflow### 1. Get the diffDefault base branch is `main`. Accept an explicit base if provided.```bashgit diff main...HEAD```If `HEAD` is already on `main` or there is no diff, fall back to staged changes:```bashgit diff --staged```If both are empty: inform the user there is nothing to review and stop.### 2. Get the current short SHA```bashgit rev-parse --short HEAD```Use this as `<sha>` in the report filename.### 3. Send to LLM for reviewCompose a prompt with the following structure:```You are a senior engineer performing a security and correctness review of a code diff.Review the diff below and produce a structured report with findings grouped by severity:- CRITICAL: security vulnerabilities, data loss, auth bypasses, path traversal, injection- HIGH: correctness bugs, panics in production paths, missing error handling, race conditions- MEDIUM: performance issues, missing cleanup, unclear invariants, incomplete implementations- LOW: style issues, dead code, missing docs, minor inefficienciesFor each finding:- Severity: CRITICAL | HIGH | MEDIUM | LOW- File + line reference (if determinable from diff)- Finding: one-sentence description- Detail: 2-3 sentences of explanation- Recommendation: concrete fixIf no findings at a severity level, omit that section.End with a one-paragraph summary verdict.DIFF:<diff content>```### 4. Write reportEnsure `.ai-logs/` directory exists, then write the report:```.ai-logs/ai-review-<sha>.md```Report format:```markdown# AI Review — <sha>**Base:** main **Date:** <date>## CRITICAL### <Finding title>**File:** `path/to/file.rs:42`**Finding:** ...**Detail:** ...**Recommendation:** ...## HIGH...## Summary<verdict paragraph>```If the directory does not exist, create it before writing.### 5. Surface resultsPrint the findings summary to the terminal grouped by severity with counts:```SEVERITY COUNTCRITICAL 0HIGH 2MEDIUM 3LOW 1Report: .ai-logs/ai-review-<sha>.md```If CRITICAL or HIGH findings exist, highlight them inline before the summary table.## Base Branch OverrideIf the user specifies a different base (e.g. "review against `next`"), use that refinstead of `main` for the diff command.## Pairing with cargo-gateRun `cargo-gate` before `ai-review` on Rust projects to ensure the diff is clean first.The review is most useful on a diff that already compiles and passes clippy.

Frequently asked questions