Where are the analysis reports saved?

Reports are automatically generated and saved in Markdown format at agentic/analysis/[type].md within your project directory for easy review.

Is the output compatible with CI/CD pipelines?

Yes, the skill returns a structured JSON object containing counts and finding details, making it ideal for integration into automated agentic workflows and CI/CD scripts.

What types of analysis can this skill perform?

It supports five specific analysis types: 'bug' for logic errors, 'debt' for architectural issues, 'doc' for documentation accuracy, 'security' for vulnerabilities, and 'style' for consistency.

Can I limit the analysis to specific folders?

Yes, you can provide optional file or directory paths as arguments to scope the analysis to specific parts of your codebase rather than the entire project.

How does the skill handle false positives?

The skill follows core principles to only report real, unfixed issues and checks if apparent problems are already handled elsewhere in the project context before flagging them.

Codebase Health Analyzer

Name: Codebase Health Analyzer
Author: e-stpierre

bye-stpierre

•

Security & Testing

Conducts multi-domain codebase audits to identify bugs, technical debt, security vulnerabilities, and documentation gaps with actionable fix suggestions.

This skill empowers Claude to perform deep-dive inspections of any codebase, categorizing findings into five critical domains: logic bugs, architectural technical debt, documentation accuracy, security vulnerabilities, and stylistic consistency. By generating both machine-readable JSON for automated workflows and human-readable Markdown reports, it provides developers with a prioritized roadmap for maintenance. The analyzer is context-aware, respecting existing project patterns, linter configurations, and CLAUDE.md guidelines to ensure high-quality, relevant findings without the noise of false positives.

Key Features

01Context-aware analysis that respects existing linter configs and project patterns

02Multi-domain scanning across bugs, debt, docs, security, and style

03Actionable fix suggestions with exact file paths and line numbers

04Dual-format output: Structured JSON and comprehensive Markdown reports

051 GitHub stars

06Four-tier severity categorization from Critical to Low impact

Use Cases

01Performing pre-release security and stability audits on feature branches

02Identifying and documenting technical debt during project handovers or refactoring phases

03Automated verification of documentation accuracy against actual code implementation

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add e-stpierre/agentic-forge analyze

For use in Claude.ai and ChatGPT

Download Skill