What is the AGENTS.md convention?

It is a standard for organizing project-specific instructions and context in a way that AI agents can easily parse and follow, using a hierarchical structure of root and scoped files to ensure the AI always has relevant, local context.

Can I customize the generated documentation?

Yes. The skill is idempotent and uses managed headers. You can manually edit the content within sections, and subsequent updates will preserve your changes while refreshing auto-extracted commands and timestamps.

How does the skill detect my project type?

The skill scans for detection signals like go.mod, package.json, composer.json, and pyproject.toml to identify languages and frameworks, automatically applying the appropriate templates for each stack.

Does this work with multi-language (hybrid) repositories?

Absolutely. The 'hybrid' mode automatically detects multiple technology stacks within a single repository and creates scoped AGENTS.md files for each directory containing significant source code (e.g., /backend and /frontend).

AGENTS.md Documentation Generator

Name: AGENTS.md Documentation Generator
Author: netresearch

bynetresearch

•

Learning & Documentation

Generates and maintains structured AGENTS.md files to provide AI agents with standardized project context, rules, and commands.

The agents skill automates the creation of hierarchical AGENTS.md documentation based on the public agents.md convention, ensuring AI assistants have the specific context they need to work effectively within your codebase. It automatically detects project stacks—including Go, PHP, TypeScript, and Python—extracts build commands from files like Makefiles or package.json, and maintains a clean separation between global project rules and subsystem-specific instructions. This tool is essential for teams looking to standardize AI onboarding, improve LLM performance on complex repositories, and maintain consistent documentation patterns across diverse software projects.

Key Features

017 GitHub stars

02Hierarchical documentation with thin root files and scoped subsystem files

03Language-specific templates for Go, PHP, TypeScript, Python, and hybrid stacks

04Auto-extraction of commands from Makefiles, package.json, composer.json, and CI workflows

05Idempotent updates that preserve custom content while refreshing auto-generated data

06Built-in validation to ensure compliance with documentation standards and precedence rules

Use Cases

01Automatically providing context and command references to Claude for faster onboarding on new codebases

02Standardizing project documentation for AI agents across a large engineering organization

03Maintaining up-to-date AI interaction patterns and safety rules in multi-repo environments

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add netresearch/claude-code-marketplace agents

For use in Claude.ai and ChatGPT

Key Features

017 GitHub stars

02Hierarchical documentation with thin root files and scoped subsystem files

03Language-specific templates for Go, PHP, TypeScript, Python, and hybrid stacks

04Auto-extraction of commands from Makefiles, package.json, composer.json, and CI workflows

05Idempotent updates that preserve custom content while refreshing auto-generated data

06Built-in validation to ensure compliance with documentation standards and precedence rules

Use Cases

01Automatically providing context and command references to Claude for faster onboarding on new codebases

02Standardizing project documentation for AI agents across a large engineering organization

03Maintaining up-to-date AI interaction patterns and safety rules in multi-repo environments