Create and maintain repo-wide AGENTS.md files for coding agents.
npx skills add https://github.com/dyxushuai/agent-skills --skill agent-init
Please help me install this Agent Skill. Skill: agent-init Source: https://github.com/dyxushuai/agent-skills/tree/main/skills/agent-init Command: npx skills add https://github.com/dyxushuai/agent-skills --skill agent-init
This skill provides a repeatable template and workflow to create and maintain repo-wide AGENTS.md files that guide how coding agents operate in a codebase.
Draft a concise AGENTS.md at the repo root with explicit commands, a repo map, change management guidance, and six core sections.
High-intent search queries and answers about installing and using this skill.
An AGENTS.md file acts as a single source of truth at the repo root to standardize how automated coding agents operate. It defines explicit commands, repo maps, and behavioral boundaries to ensure consistent agent actions across teams.
To create an AGENTS.md, draft a concise root document using a repeatable template. Include explicit run and test commands, a repo map, change management guidance, and six core sections covering always, ask-first, and never behavioral boundaries.
Yes, an AGENTS.md ensures cross-agent compatibility for platforms like Claude Code and Copilot. It provides specific compatibility notes and symlink guidance to maintain consistent agent behavior across different automated tools.
Yes, this approach targets single-repo or small monorepos. A single root AGENTS.md document provides consistent agent behavior, auditing capabilities, and onboarding guidance across multiple teams and agent platforms.
A root AGENTS.md is designed for single-repo or small monorepos. It may not scale effectively for large, complex monorepos requiring highly granular, directory-specific agent instructions, as it focuses on maintaining a single source of truth.
Standardizing repo agent behavior requires defining always, ask-first, and never boundaries in your AGENTS.md. This ensures automated agents understand permitted actions, restricted operations, and mandatory approval workflows before executing changes.
RheinmirIdentify repository structure, entrypoints, dependency flow, and build/run/test paths.
GaryOcean428Synchronize AGENTS.md and CLAUDE.md files across agent platforms via symlinks.
rodmena-limitedCreate, configure, and manage Claude Code Skills with templates and tools.
markjoshwelAuthor concise AGENTS.md files using six-core-areas and three-tier boundaries.
