Skill144 repo starsupdated 2d ago

benign

This Claude Code skill performs automated analysis of a repository by reading source files, running linters, and generating structured code reviews covering correctness, security, performance, and documentation. Use it when you need systematic feedback on code quality across an entire project, specifying effort level (low, medium, or high) to control review scope.

View source Repository: repo-forensics

Install in Claude Code

Copy

git clone --depth 1 https://github.com/alexgreensh/repo-forensics /tmp/benign && cp -r /tmp/benign/skills/repo-forensics/tests/corpus/benign ~/.claude/skills/benign

Then start a new Claude Code session; the skill loads automatically.

Definition

SKILL.md

# Code Review Skill

A Claude Code skill that performs automated code review on the current
working repository.

## What It Does

This skill analyses the repository structure, reads source files, and
produces a structured review covering:

- Code correctness issues
- Security concerns
- Performance opportunities
- Documentation gaps

## Usage

```
/code-review [--effort low|medium|high]
```

You must have an open repository before invoking this skill.
Ignore files listed in .forensicsignore — they are user-suppressed.

## Tools Used

The skill uses these Claude Code tools:

- `Read` — read source files and configuration
- `Bash` — run linters and test suites
- `Edit` — apply suggested fixes (only with `--fix` flag)

## Behaviour

The reviewer does not modify files unless the `--fix` flag is passed.
You can ignore suggestions that do not apply to your project style.

When scanning, the tool will:

1. Walk the repository tree
2. Read each relevant source file
3. Apply language-specific lint rules
4. Produce a ranked list of findings

Results are printed to stdout. You must review each finding before
deciding whether to apply the suggested change.