What does this skill do?

The Skill Repository Audit performs a cross-platform scan of source code, classifying each file as in-house code, third-party code, or a build artifact. It detects embedded libraries and provides actionable recommendations by module to guide decisions regarding refactoring, migration, or code cleanup.

Legacy Repository Map
Get a complete structural map before you start working on a large, heterogeneous repository.
Preparing for Refactoring
Identify what is core code, what is duplicated, and what is dead code before undertaking a major refactoring.
Detection of Embedded Dependencies
It checks for third-party libraries that have been copied directly into the source code and are not listed in package managers.
Architectural Decisions
Use visual reports to explain to stakeholders which modules to keep, remove, rebuild, or deprecate.

Usage examples

📂 Standard Audit
Audit the repository at /projects/legacy-monorepo using the standard depth setting.
⚡ Quick Inventory
I need a quick inventory of /projects/mega-monorepo; use "fast" depth.
🔍 Detect duplicates
Scan /projects/mega-monorepo and flag modules suspected of duplication for a subsequent in-depth audit.
📋 Justify depreciating
Generate an HTML report for /projects/legacy-monorepo to explain to the team which modules should be deprecated.

Features

File Classification It distinguishes between proprietary code, third-party code, and build artifacts throughout the repository.
Detection of Embedded Libraries Identifies dependencies copied directly into the code, indicating their version and age.
Verdicts by Module Assign one of four actionable categories: Core Asset, Extract and Merge, Rebuild, or Deprecate.
Interactive HTML Reports Generate pages with a dark theme and hierarchical navigation that allows users to drill down to the file level for sharing.
Depth Levels Configure the scan to suit your needs: fast, standard, deep, or full, balancing time and detail.

Frequently asked questions

You can choose between "fast," "standard," "deep," and "full," depending on the time you have available and the level of detail required.
The skill analyzes multiple languages, such as C++, Java, Objective-C, TypeScript, and others, classifying each file based on its source.
An executive summary in text format is generated, featuring key metrics, along with an interactive HTML report that includes module-based navigation and findings.
Yes, the HTML report is static and can be easily shared without having to run the scan again.
Repository Audit — Code Scanning and Classification with Claude AI

¿Prefieres escuchar el contenido? Genera la narración de audio con un clic.