Does it work with monorepos?

Yes, the tool automatically detects monorepo structures and allows you to focus the analysis on specific packages or the entire root workspace.

How does this help Claude subagents?

It provides a structured 'knowledge profile' that allows subagents to understand project-specific patterns and architectural decisions without scanning the entire repo every time, leading to more accurate code generation.

Where are the profiles stored?

Profiles are stored globally in ~/.claude/knowledge/ and symlinked locally to your project's .agent/local/ directory for easy access and version control compatibility.

What is the difference between Quick and Deep analysis?

Quick analysis provides a ~2,000-word overview for immediate tasks, while Deep analysis samples more files and traces complex dependency chains to produce a ~6,000-word expert profile.

Subagent Knowledge Profile Generator

Name: Subagent Knowledge Profile Generator
Author: hideki5123

byhideki5123

•

Learning & Documentation

Generates deep codebase knowledge profiles to provide AI subagents with instant domain expertise and architectural context.

This skill automates the creation of a comprehensive PROJECT-KNOWLEDGE.md file by performing a multi-dimensional analysis of your codebase. It scans directory structures, architecture, coding conventions, and domain concepts using parallel subagents, outputting a persistent profile that can be loaded into future AI prompts. This ensures that any AI agent working on the project has the context of an expert teammate, reducing hallucinations and improving the accuracy of complex code modifications.

Key Features

011 GitHub stars

02Smart monorepo support and incremental profile updates

03Persistent global and local knowledge storage for AI context

04Multi-dimensional codebase scanning (Structure, Architecture, Conventions, Domain)

05Configurable analysis depth ranging from Quick to Deep expert levels

06Automated tech stack and dependency detection

Use Cases

01Maintaining an up-to-date architectural map for automated subagent prompts

02Onboarding an AI agent to a complex, legacy, or unfamiliar codebase

03Creating a standardized 'Context Pack' for team-wide AI development collaboration

Key Features

011 GitHub stars

02Smart monorepo support and incremental profile updates

03Persistent global and local knowledge storage for AI context

04Multi-dimensional codebase scanning (Structure, Architecture, Conventions, Domain)

05Configurable analysis depth ranging from Quick to Deep expert levels

06Automated tech stack and dependency detection

Use Cases

01Maintaining an up-to-date architectural map for automated subagent prompts

02Onboarding an AI agent to a complex, legacy, or unfamiliar codebase

03Creating a standardized 'Context Pack' for team-wide AI development collaboration