# Review code with vet after every change

Vet helps AI coding assistants identify issues in their code changes by analyzing git diffs and conversation history. Run it immediately after any logical unit of code changes to catch misunderstandings between what was requested and what was implemented.

## Install

```bash
npx skillstore add imbue-ai/vet
```

## Metadata

- - Slug: imbue-ai-vet
- - Version: 1.0.0
- - Author: imbue-ai
- - GitHub username: imbue-ai
- - License: MIT
- - Repository: https://github.com/imbue-ai/vet/tree/main/skills/vet
- - Ref: main
- - Supported tools: Claude, Codex, Claude Code
- - Risk level: safe
- - Quality score: 72
- - Public page: https://skillstore.pages.dev/skills/imbue-ai-vet
- - Manifest: https://skillstore.pages.dev/api/skills/imbue-ai-vet/manifest

## Capabilities

- Analyzes git diffs to find issues in code changes
- Reviews conversation history to catch misunderstandings between user requests and implementation
- Works with Claude Code, Codex, and OpenCode session formats
- Supports configurable confidence thresholds for findings
- Outputs results in text, JSON, or GitHub-compatible formats
- Can route analysis through local CLI instead of API for offline use

## Use Cases

- Catch bugs early in development: Run vet after each code change to catch bugs before they accumulate. The tool reviews the git diff and conversation context to identify potential issues.
- Improve code quality during feature development: Use vet to verify implementation matches user intent by analyzing conversation history alongside code changes.
- Review AI-generated code manually: Vet provides an additional layer of review for AI-generated code, catching issues that the AI may have missed.

## Prompt Templates

### Basic vet check

```
Run vet on the current changes to review the code.
```

### Vet with specific focus

```
Run vet on the current changes focusing on security issues.
```

### Vet with custom model

```
Run vet using the claude-sonnet-4 model with a 0.9 confidence threshold.
```

### Vet with agentic mode

```
Run vet in agentic mode to route through local Claude Code CLI.
```

## Limitations

- Does not replace running tests - it is a complementary review tool
- May report issues from other agents or sessions in shared repositories
- Requires access to session history files to function with full context
- Requires Python with pip or uv to install the vet package

## Best Practices

- Run vet immediately after every logical unit of code changes, not in batches
- Review issues that relate to changes you actually made, not other agents
- Use the --agentic mode when API keys are unavailable but expect slower analysis

## Anti Patterns

- Do not ignore issues without reviewing them - they may indicate real problems
- Do not run vet only at the end of a large feature - catch issues early
- Do not assume vet replaces testing - it complements but does not replace tests

## Security Audit

- - Safe to publish: true
- - Audited at: 2026-02-22T05:14:22.534\+00:00
- - Summary: This is a legitimate code review tool that analyzes AI assistant session history and git diffs. The static scanner flagged 54 potential issues, but all are false positives. The 'weak cryptographic algorithm' alerts are triggered by argparse \(command-line parsing\), not crypto code. 'Shell backtick' alerts are markdown code blocks showing example commands, not actual execution. 'Hidden file' and 'system reconnaissance' alerts are legitimate file path accesses to session storage required for the tool to function.

## Stats

- - Views: 104
- - Downloads: 3
- - Favorites: 0
- - Popularity score: 0