Brutal Codebase Audit FAQs

Question 1

Can I audit just my most recent changes?

Accepted Answer

Yes, the --recent flag scopes the audit specifically to files changed within the last 20 commits, allowing for faster feedback loops during active development.

Question 2

Is it safe to let the agent apply fixes automatically?

Accepted Answer

While the agent validates fixes before applying them, it is recommended to run the audit on a clean git branch so you can review and test the changes before merging them into your main codebase.

Question 3

Which Claude models does this skill use?

Accepted Answer

The skill utilizes Claude 3 Opus for high-level architectural and logic analysis, while leveraging Claude 3.5 Sonnet for security checks, convention compliance, and fix validation.

Question 4

How does the audit skill handle false positives?

Accepted Answer

It uses a two-stage process where initial findings from scanning agents are passed to a separate validation agent that confirms the issue and the proposed fix before it is reported or applied.

Question 5

Can I run this audit without modifying my files?

Accepted Answer

Yes, by using the --dry-run argument, the skill will generate a full report of findings and proposed fixes without making any changes to your source code.

Brutal Codebase Audit

Key Features

Use Cases

Brutal Codebase Audit

Key Features

Use Cases