Skip to content

AGENTS.md

Latest commit

 

History

History
70 lines (55 loc) · 2.75 KB

AGENTS.md

File metadata and controls

70 lines (55 loc) · 2.75 KB

Agent Skills Repository

A collection of reusable agent skills packaged in the Agent Skills format. Skills extend AI coding agents with domain-specific knowledge and instructions.

Tech Stack

  • Markdown (SKILL.md files with YAML frontmatter)
  • Bash scripts (optional, in scripts/ directories)
  • Distributed via npx skills CLI (skills.sh)

Repository Structure

skills/
  └── <skill-name>/            # kebab-case, matches `name` in frontmatter
      ├── SKILL.md             # Required. Frontmatter + instructions (<500 lines)
      ├── README.md            # Optional. Human-readable docs, install instructions
      ├── references/          # Optional. Deep technical content loaded on-demand
      │   └── REFERENCE.md
      ├── scripts/             # Optional. Executable scripts agents can run
      └── assets/              # Optional. Templates, data files, static resources

Commands

Command Purpose
npx skills add notcodev/agent-skills Install all skills from this repo
npx skills add notcodev/agent-skills --skill <name> Install a specific skill
npx skills add notcodev/agent-skills --list Preview available skills

Creating a New Skill

  1. Create skills/<skill-name>/SKILL.md with YAML frontmatter: 
    ---
name: skill-name
description: >
  What it does and when to trigger.
  TRIGGER when: <conditions>.
  DO NOT TRIGGER when: <conditions>.
license: Apache-2.0
metadata:
  author: notcodev
  version: "1.0"
---
  2. Write instructions in Markdown after the frontmatter. Keep under 500 lines.
  3. Put detailed code examples and deep reference material in references/REFERENCE.md.
  4. Add a README.md with human-readable docs and npx skills add install commands.
  5. Update the root README.md skills table.

Skill Authoring Rules

  • The name field must be kebab-case, match the directory name, 1-64 characters.
  • The description field must include trigger conditions so agents know when to activate.
  • SKILL.md body should be concise instructions and rules. Move lengthy code examples to references/.
  • Scripts in scripts/ must use set -e and output JSON when returning structured data.
  • Never put secrets, API keys, or credentials in skill files.

Code Style

  • Markdown files: standard CommonMark with GFM tables and fenced code blocks.
  • YAML frontmatter: 2-space indent, quoted strings for versions.
  • Code examples in skills: use TypeScript with explicit types, no any unless necessary.

Boundaries

  • Do not add ORM-specific or framework-specific instructions outside of the relevant skill context.
  • Each skill must be self-contained within its directory.