AGENTS.md
The open format for a README addressed to agents: one Markdown file at the repository root, nested files allowed, read by Codex, Cursor, Copilot, Gemini CLI, Grok Build and others.
Before an agent loads a skill it reads the instruction files in your repository: AGENTS.md, CLAUDE.md, Cursor rules, Copilot instructions. These collections hold reviewed examples, templates and the standards that define the formats.
AGENTS.md is a plain Markdown file at the repository root that most coding agents read, including Codex, Cursor, Copilot, Gemini CLI, Grok Build and Claude Code when no CLAUDE.md is present. Cursor adds project rules in .cursor/rules as .mdc files with scoping frontmatter; Copilot reads .github/copilot-instructions.md and path-specific instruction files.
Persona files such as SOUL.md and prompt libraries such as Fabric's patterns are the same idea for assistants that live in chat apps: text the agent loads every session. None of these carry code, which makes them the easiest layer to review before you adopt them.
Copy the file into the place your agent reads. Most of them need no tooling: AGENTS.md goes at the repository root, Cursor rules into .cursor/rules, Copilot instructions into .github, CLAUDE.md beside your code. Plugins for Codex, Cursor and Gemini CLI install with each product's own command.
codex plugin marketplace add owner/repo
gemini extensions install https://github.com/gemini-cli-extensions/workspace
copilot plugin marketplace add github/awesome-copilot
Documentation: AGENTS.md · Cursor rules · Copilot repository instructions · Claude Code memory and CLAUDE.md · Codex skills · Gemini CLI extensions · Agent Skills specification · Agent Plugins standard
AGENTS.md, CLAUDE.md, Cursor rules and Copilot instructions.
The open format for a README addressed to agents: one Markdown file at the repository root, nested files allowed, read by Codex, Cursor, Copilot, Gemini CLI, Grok Build and others.
Cursor rule files by framework and language, as .mdc files to copy into .cursor/rules.
Community agents, instruction files, skills, hooks and plugins for GitHub Copilot, installable as a Copilot plugin marketplace.
copilot plugin marketplace add github/awesome-copilot
Exemplary CLAUDE.md files collected from public projects.
CLAUDE.md templates by stack, with rules and workflow files.
The other agents' plugin formats and their official catalogs.
The Cursor plugin specification and the official plugins, each a .cursor-plugin folder with rules, skills, agents, commands, MCP servers and hooks.
The catalog behind cursor.directory, where community plugins and MCP servers are submitted by repository URL.
Curated Codex plugin examples, including Figma, Notion, Expo, Netlify, Remotion and Google Slides.
codex plugin marketplace add openai/plugins
Codex's system, curated and experimental skills. The repository now points new work at openai/plugins, but its curated skills still install by name with the skill installer.
A spec-driven development extension for Gemini CLI, from Google's extensions organization.
gemini extensions install https://github.com/gemini-cli-extensions/conductor
Google Workspace access for Gemini CLI, packaged as an extension with its MCP servers and commands.
gemini extensions install https://github.com/gemini-cli-extensions/workspace
Text the agent loads every session: who it is and how it writes.
SOUL.md persona templates for OpenClaw, the self-hosted assistant that lives in chat apps.
An open-source framework of reusable prompt patterns, each a system.md file, run from the command line.
fabric -p <pattern>
The largest open prompt library, published as a CSV with a Claude Code plugin and an MCP server.
xAI's published system prompts for Grok.
What defines a skill, a plugin and the installer most agents share.
The open standard behind every skill in this directory: a SKILL.md with name and description frontmatter, optional scripts, references and assets, and a validator.
A vendor-neutral standard for packaging skills, MCP configuration and extensions into one portable plugin folder.
The installer behind npx skills add: fetches a skill folder from GitHub or another source and places it in the directory each of more than 75 agents reads.
npx skills add owner/repo -a claude-code
Entries are added by hand from their publisher's repository, checked to exist on November 2, 2026, and described from what the publisher documents. They carry no Suede Skill Check score, no security result and no verification badge: the score applies to SKILL.md skills only. To suggest a repository or a correction, open a listing request.
A plain Markdown file at the repository root with instructions for coding agents, stewarded by the Agentic AI Foundation. Codex, Cursor, Copilot, Gemini CLI and Grok Build read it, and Claude Code reads it when the project has no CLAUDE.md.
Project rules live in .cursor/rules as .mdc files with description, globs and alwaysApply frontmatter. A plain .md file in that folder is ignored. Cursor also reads AGENTS.md at the root and in subfolders.
No. Instruction files, rules and prompts are text the agent reads. That makes them the easiest layer to review, but a file can still steer an agent toward risky actions, so read it before you adopt it.
No. They are listed from their publisher's repository with a short description and carry no score, security result or verification badge. The Suede Skill Check applies to skills only.