agent-init
🎯Skillfrom dyxushuai/agent-skills
|Creates or improAGENTS.repository configuration file AGENTS.with md) with standardized build, test,,, boundaries, and symlinklink support for multi-agent workflows across across repositories. Hum...
Installation
npx skills add https://github.com/dyxushuai/agent-skills --skill agent-initSkill Details
Overview
# Agent Skills
A growing set of skills for consistent, repo-aware agent workflows.
Skills follow the Agent Skills specification: https://agentskills.io
Available Skills
agent-init
Create or improve AGENTS.md files that define how coding agents operate in a repo.
Designed to replace ad hoc "init" prompts with a consistent, repo-aware workflow.
Reference (best practices + starter template):
https://github.blog/ai-and-ml/github-copilot/how-to-write-a-great-agents-md-lessons-from-over-2500-repositories/#starter-template
Use when:
- Setting up an initial AGENTS.md
- Replacing an agent "init" command with a durable AGENTS.md
- Standardizing multi-agent behavior across a repo
- Auditing an existing AGENTS.md for missing commands or boundaries
- Using Claude Code (requires CLAUDE.md symlink to AGENTS.md)
Location: skills/agent-init
Installation
Install all skills from this repo:
```bash
npx add-skill dyxushuai/agent-skills
```
Install only agent-init:
```bash
npx add-skill dyxushuai/agent-skills --skill agent-init
```
List skills:
```bash
npx add-skill dyxushuai/agent-skills --list
```
Best Practices Summary
Follow GitHub's guidance: put commands early (with flags), include at least one real code example, specify tech stack versions, and use Always / Ask first / Never boundaries; keep AGENTS.md short, concrete, and repo-specific.
Usage
Skills are available once installed. The agent will use them when the request matches.
Codex example
```
$agent-init
Create AGENTS.md with exact build/test/lint commands and boundaries.
```
Claude example
```
/agent-init
Create AGENTS.md with exact build/test/lint commands and boundaries.
Create CLAUDE.md as a symlink to AGENTS.md.
```
Other prompts
- "Audit our existing AGENTS.md for missing boundaries and commands."
- "Update AGENTS.md to include the current test and lint commands."
- "Add CLAUDE.md as a symlink to AGENTS.md for Claude Code."
Skill Structure
Each skill contains:
SKILL.md- instructions for the agentreferences/- supporting documentation (optional)scripts/- helper scripts (optional)
Compatibility
Skills are plain Markdown with YAML frontmatter and work across multiple agents.
Install with add-skill for Codex, Claude, Cursor, and others.
If Claude Code is in scope, create CLAUDE.md as a symlink to AGENTS.md.