Why AgentSync?

Single Source of Truth

Maintain one AGENTS.md and sync it to Claude, Copilot, Gemini, and more via symlinks.

Native Symlinks

Uses symbolic links instead of copies. Change a file once, and every assistant sees it immediately.

Works Across Assistants

Configure many assistant targets; use native MCP formatting for the supported assistants that have a formatter.

MCP Integration

Define MCP servers once in TOML and sync them across all supported agent environments.

Skills

Install and manage reusable skills (from skills.sh or local archives) under .agents/skills/.

Gitignore Management

Start with the default managed .gitignore workflow, or opt out intentionally when your team commits managed destinations.

Cross-Platform

Written in Rust for maximum speed and portability. Works on macOS, Linux, and Windows.

How AgentSync works

1. 01

   Define

   Keep instructions, skills, and MCP configuration in .agents/ — the single source of truth.

   .agents/AGENTS.md

2. 02

   Apply

   Run one command. AgentSync materializes native symlinks — no copies, no drift.

   $ agentsync apply
   ✔ Linked: CLAUDE.md -> .agents/AGENTS.md

3. 03

   Synced

   Every assistant reads the same files. Verify the state anytime.

   ✔ OK: CLAUDE.md -> .agents/AGENTS.md
   Status: All good

---

Pick a path

Start with the path that matches your goal:

• Install and sync your first assistant: follow Getting Started.
• Configure common setups: follow the Configuration Recipes.
• Organize instruction files: read Instruction Files and Sync Types.
• Configure MCP servers or skills: use the MCP Integration and Skills guides.
• Check intended configuration or drift: use Inspect Configuration and Status Output.
• Automate team setup or diagnose an issue: see Git Hook Automation, Gitignore Team Workflows, and Troubleshooting.

For field definitions and command flags, use the Configuration Reference and CLI Reference.