Can this skill handle monorepos?

Yes, it supports nested configurations, allowing each package to have its own specific AGENTS.md while maintaining a lean root file for workspace-wide settings.

Why should I use AGENTS.md instead of CLAUDE.md?

AGENTS.md is an open standard compatible with over 20 AI tools like Cursor and Windsurf. This skill maintains AGENTS.md as the source of truth and symlinks CLAUDE.md to it for Claude-specific compatibility.

What is progressive disclosure in this context?

It's the practice of keeping the root configuration file minimal and linking to domain-specific files only when needed, preventing the agent from being overwhelmed by irrelevant instructions on every request.

How does this improve agent performance?

By staying within an 'instruction budget' of 150-200 rules, the agent follows instructions more consistently and consumes fewer tokens, leading to faster and more accurate responses.

AGENTS.md Config Architect

Name: AGENTS.md Config Architect
Author: richtabor

byrichtabor

•

콘텐츠 관리

Standardizes agent configuration files using progressive disclosure to optimize LLM performance and context efficiency.

This skill empowers Claude to create, refactor, and maintain minimal, high-performance AGENTS.md files and symlinked CLAUDE.md files following industry best practices. By prioritizing progressive disclosure over exhaustive documentation, it prevents context poisoning from stale data and keeps agent instructions within an efficient token budget. It proactively identifies missing configurations and helps migrate bloated documentation into a structured, multi-file hierarchy that scales with complex projects and monorepos, ensuring your AI tools always have the most relevant information without the noise.

주요 기능

0141 GitHub stars

02Anti-pattern detection for stale file trees and redundant rules

03Automated AGENTS.md creation and CLAUDE.md symlinking

04Monorepo-aware configuration nesting for scaled projects

05Bloat reduction via progressive disclosure principles

06Token-efficient project summaries and command tracking

사용 사례

01Refactoring a bloated CLAUDE.md into a lean AGENTS.md hierarchy

02Initializing a new project with standardized, cross-tool agent configurations

03Optimizing LLM instruction following by reducing instruction noise

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add richtabor/agent-skills review-agents-md

For use in Claude.ai and ChatGPT

Download Skill