When should I activate this skill?

The skill activates automatically when you ask Claude to 'understand', 'investigate', or 'research' a specific subsystem or domain within your codebase.

How does the skill decide what information is worth recording?

The skill follows a strict protocol that prioritizes generalizable patterns, non-obvious conventions, and architectural 'gotchas' over specific implementation details that only apply to a single file.

Will this skill create a file in every folder of my project?

No. The skill follows placement rules that only trigger the creation of a new AGENTS.md file if a domain has 3+ files, distinct patterns, or significant complexity that warrants isolation.

Does it overwrite my existing documentation?

No, the skill is designed to append to existing sections or create new ones, ensuring that previous learnings are preserved while new discoveries are integrated.

What is an AGENTS.md file?

An AGENTS.md file is a localized documentation file used by this skill to store specific patterns, conventions, and warnings relevant to a particular directory or subsystem.

Research Compounder

Name: Research Compounder
Author: majesticlabs-dev

bymajesticlabs-dev

•

학습 및 문서화

Automates the capture and retrieval of domain-specific architectural context and patterns during codebase research.

The Research Compounder skill transforms how Claude interacts with complex repositories by institutionalizing knowledge through localized AGENTS.md files. It systematically traverses folder hierarchies to extract existing context before research begins, ensuring Claude adheres to established conventions. After investigation, it automatically appends newly discovered patterns, gotchas, and testing strategies to the relevant directory's documentation. This creates a self-documenting ecosystem that prevents redundant discovery efforts and maintains high-quality coding standards across diverse subsystems like jobs, services, and controllers.

주요 기능

01Automatic AGENTS.md generation for new or complex sub-domains

02Structured documentation updates for Gotchas, Patterns, and Testing

03Hierarchical context retrieval using a nearest-wins search for AGENTS.md

04Smart filtering to distinguish generalizable patterns from one-off details

0518 GitHub stars

06Batch processing of discoveries to ensure clean documentation history

사용 사례

01Maintaining architectural consistency across microservices or multi-domain applications

02Onboarding to a large-scale legacy codebase with complex, undocumented conventions

03Preventing recurring bugs by documenting non-obvious 'gotchas' during subsystem research

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add majesticlabs-dev/majestic-marketplace research-compound

For use in Claude.ai and ChatGPT

Download Skill