Can I audit code running in a remote or containerized environment?

Yes. The skill supports a 'Paste output' method, allowing you to run the instrumented code in any environment and bring the logs back to Claude for analysis.

What programming languages does the Feature Audit skill support?

It provides out-of-the-box support for TypeScript, JavaScript, Python, Go, Rust, and Java using language-appropriate logging patterns and markers.

Where are the audit reports and logs saved?

All artifacts, including the final report, session metadata, and captured logs, are stored in a structured directory at docs/audits/[audit-id]/.

Will this skill permanently change my source code?

No. While the skill injects logs for auditing, it tracks every change in a manifest file and includes a dedicated cleanup phase to restore your code to its exact original state.

How does this differ from standard static code analysis?

Static analysis infers what code should do by reading it; Runtime Audit proves what code actually does by observing it during execution with injected instrumentation.

Runtime Feature Audit

Name: Runtime Feature Audit
Author: schuettc

byschuettc

•

Analytics & Monitoring

Verifies code behavior through automated log injection and systematic runtime execution analysis.

The Feature Audit skill bridges the gap between static code analysis and actual runtime behavior by orchestrating a systematic auditing workflow. It automatically identifies strategic execution points, safely injects language-appropriate log markers, captures real-time data from process execution, and generates comprehensive reports. Designed with a "trust but verify" philosophy, it includes a robust rollback mechanism that tracks every modification in a manifest file, ensuring code is perfectly restored to its original state after the audit is complete. This is an essential tool for debugging complex flows, verifying feature implementation, and understanding legacy code behavior without permanent codebase pollution.

Key Features

01Safe rollback guarantee using an injections.json manifest for 100% code restoration.

02Automated log injection for multiple languages including TypeScript, Python, Go, Rust, and Java.

03Systematic 7-phase workflow from target identification to evidence-based verification.

04Persistent audit reporting with structured documentation of verified vs. unexpected behaviors.

05Flexible runtime capture supporting both direct bash execution and external log pasting.

064 GitHub stars

Use Cases

01Creating verifiable documentation of system behavior for compliance or architectural reviews.

02Debugging complex, non-deterministic runtime issues that static analysis cannot identify.

03Verifying that a newly implemented feature follows the intended execution path and data flow.

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add schuettc/claude-code-plugins feature-audit

For use in Claude.ai and ChatGPT

Download Skill