Does this skill work on small or single-file projects?

Analysis is considered premature for projects with fewer than 5 files or no application code; in these cases, it falls back to basic stack detection.

How does the skill select files for deep analysis?

It uses specific sampling criteria: entry points, modules with high import counts, domain-keyword paths, and recently active files from git history.

How does codebase analysis differ from stack detection?

Stack detection identifies WHAT technologies exist in a project, whereas codebase analysis evaluates HOW they are used and WHETHER they warrant specialized agents or skills.

Which files are considered high-signal for analysis?

The skill prioritizes CLAUDE.md as the single source of truth for conventions, followed by README.md and project-level identity documentation.

What triggers a specialist agent recommendation?

Specialists are suggested based on signals like background job directories, payment models, ML scripts, or custom middleware layers with multiple implementations.

Codebase Analysis Workflow

Name: Codebase Analysis Workflow
Author: Kiwi-Home

byKiwi-Home

0•

开发者工具

Analyzes project structure, domain significance, and coding patterns to determine when specialized agents or skills are required.

This skill provides a rigorous framework for Claude to evaluate any codebase by identifying high-signal documents like CLAUDE.md and READMEs, sampling critical entry points, and detecting domain-specific indicators. It enables AI agents to make informed decisions about project architecture, identifying where custom automation is warranted based on integration breadth, risk profiles, and development activity. By formalizing the analysis of conventions and specialist signals, it ensures that generated agents and skills are aligned with the actual needs and patterns of the project rather than framework defaults.

主要功能

01Hierarchical signal evaluation prioritizing project identity documents

02Intelligent file sampling based on entry points and import density

03Automated identification of specialist agent and skill requirements

04Staleness detection for maintaining generated asset relevance

05Domain significance scoring using risk and complexity metrics

060 GitHub stars

使用场景

01Extracting coding conventions to prepopulate project-specific documentation and rules

02Determining if a project needs specialized sub-agents for areas like payments or ML

03Onboarding an AI agent to a new, complex repository to establish a mental model

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add kiwi-home/ai-coding-resources codebase-analysis

For use in Claude.ai and ChatGPT

Download Skill