# Agsync

> Agsync is a repo-native control plane for managing skills, MCP servers, secrets, and agent configs across AI clients like Claude Code and Cursor.

- **Type:** MCP server
- **Install:** `agentstack add mcp-yiftahb-agsync`
- **Verified:** Yes — security-reviewed for prompt injection and unsafe behavior
- **Seller:** [yiftahb](https://agentstack.voostack.com/s/yiftahb)
- **Installs:** 0
- **Category:** [Security](https://agentstack.voostack.com/c/security)
- **Latest version:** 0.1.0
- **License:** MIT
- **Upstream author:** [yiftahb](https://github.com/yiftahb)
- **Source:** https://github.com/yiftahb/agsync

## Install

```sh
agentstack add mcp-yiftahb-agsync
```

Requires the [AgentStack CLI](https://agentstack.voostack.com/docs/cli). Works with Claude Code, Cursor, and any MCP-compatible agent.

## About

Git-native CLI to sync skills, commands, and MCP servers across AI coding agents.

  
  
  
  

## Supported Agents

| Agent | Instructions | Skills | Commands | MCP |
|-------|-------------|--------|----------|-----|
| Claude Code | `CLAUDE.md` | `.claude/skills/` | `.claude/commands/` | `.mcp.json` |
| Cursor | `AGENTS.md` | `.cursor/skills/` | `.cursor/commands/` | `.cursor/mcp.json` |
| Codex | `.codex/instructions.md` | `.codex/skills/` | `.codex/commands/` | `.codex/config.toml` |
| Windsurf | `AGENTS.md` | `.windsurf/skills/` | `.windsurf/commands/` | `.windsurf/mcp_config.json` |
| Copilot | `.github/copilot-instructions.md` | -- | `.github/agents/` | `.mcp.json` |
| Gemini | `GEMINI.md` | `.gemini/skills/` | `.gemini/commands/` | `.gemini/settings.json` |
| OpenCode | `AGENTS.md` | `.opencode/skills/` | `.opencode/command/` | `opencode.json` |
| Antigravity | `.agent/rules/instructions.md` | `.agent/skills/` | `.agent/commands/` | -- |

## Install

```bash
npm i -g agsync-cli
```

Or run directly:

```bash
npx agsync-cli
```

## Features

🤖 **Multi-Agent Sync** — Define skills, commands, and instructions once in `.agsync/`, generate native output for every agent your team uses. Symlinks keep everything in sync.

🧩 **Skill Extension** — Import skills from GitHub or ClawHub, extend them with local overrides, and resolve inheritance chains automatically. Supporting files (rules, scripts, references) are bundled alongside.

🔒 **Version Locking** — Pin external skills to exact versions. A lock file (`agsync-lock.yaml`) tracks resolved commits and content hashes for reproducible builds. Use `--frozen` in CI to enforce it.

🔌 **MCP Sync** — Define MCP servers in YAML, expand environment variables at sync time, and generate the correct config format (JSON or TOML) per agent with smart merging.

🏷️ **MCP Namespaces** — Tag MCP servers with namespaces (e.g. `ci-cd`, `coding-agents`) and filter at sync time with `agsync sync --namespace ` so each environment only gets the servers it needs.

📁 **Monorepo Scoping** — One `agsync.yaml` at the repo root, with `.agsync/` directories in any subfolder. Skills and commands are automatically prefixed (e.g. `frontend:my-skill`), built to the root, and each subfolder gets its own `AGENTS.md` with scoped instructions.

🔒 **Gitignore Management** — Automatically manage `.gitignore` entries for generated output. Choose between `on` (all output), `mcpOnly` (default, MCP configs only), or `off`.

📐 **Structured Context** — Enable `features.context: true` to write `.agsync/instructions.md` with structured `## HLD`, `## Guidelines`, and `## Patterns` sections. Guidelines cascade through the full nested hierarchy (root → `backend` → `backend/service-a`), with nearest-scope overrides by `id`. Paths resolve relative to the scope that declares them, so one shared pattern applied in several services scopes to each (`backend/controllers/**`, `service-a/controllers/**`). Each subfolder compiles a self-contained `AGENTS.md` carrying its full inherited ruleset.

🧾 **Code Review Output** — Enable `review: targets: [coderabbit]` to generate `.coderabbit.yaml` at the repo root on every `agsync sync`. Guidelines and their path scopes map directly to coderabbit `path_instructions` — one source of truth for agents and reviewers alike.

## Commands

| Command | Description |
|---------|-------------|
| `agsync init` | Scaffold a new project with `agsync.yaml` and `.agsync/` |
| `agsync skill add ` | Create a local empty skill |
| `agsync skill add github:` | Import a skill from GitHub |
| `agsync skill add clawhub:` | Import a skill from ClawHub |
| `agsync skill remove ` | Remove a skill |
| `agsync command add ` | Create a new command (.md) |
| `agsync command remove ` | Remove a command |
| `agsync mcp add ` | Create a new tool definition (.yaml) |
| `agsync mcp remove ` | Remove a tool |
| `agsync validate` | Validate config, skills, commands, and tool references |
| `agsync plan [--frozen] [--namespace ]` | Preview changes without writing files |
| `agsync sync [--frozen] [--namespace ]` | Generate output for all enabled agents |
| `agsync doctor` | Check environment health and enabled agents |
| `agsync version` | Show current version and check for updates |
| `agsync update` | Update to the latest version |
| `agsync help` | Show extended help |

## Monorepo Support

One `agsync.yaml` at the repo root, multiple `.agsync/` directories in subfolders:

```
project/
├── agsync.yaml              ← single config
├── .agsync/                 ← root skills, commands, mcp
├── frontend/
│   └── .agsync/             ← frontend-specific definitions
│       ├── instructions.md  → generates frontend/AGENTS.md
│       └── skills/ui-kit/
└── backend/
    └── .agsync/
        └── skills/api/
```

- Scoped skills are prefixed automatically (`frontend:ui-kit`) and output to `.agents/skills/frontend--ui-kit/` at the root
- Each subfolder with `.agsync/instructions.md` gets its own `AGENTS.md` generated in-place
- The root `AGENTS.md` cross-references scoped instructions: *"When working in folder: `frontend` — you MUST load `frontend/AGENTS.md`"*
- Agent symlinks (`CLAUDE.md`, `.claude/skills/`) are only created at the root
- `agsync skill add`, `command add`, `mcp add` resolve the nearest `.agsync/` upwards, so running from `frontend/` adds to `frontend/.agsync/`

## MCP Namespaces

Tag MCP servers with one or more namespaces and filter at sync time. Useful when CI runners and local coding agents need different sets of servers.

```yaml
# .agsync/mcp/github-actions.yaml
name: github-actions
description: GitHub Actions MCP server
namespaces: [ci-cd]
command: npx
args: ["-y", "@modelcontextprotocol/server-github-actions"]
```

```yaml
# .agsync/mcp/cursor-tools.yaml
name: cursor-tools
description: Tools for the local coding agent
namespaces: [coding-agents]
command: npx
args: ["-y", "@some/cursor-tools"]
```

Sync only the servers tagged for a namespace:

```bash
agsync sync --namespace ci-cd          # only ci-cd-tagged + untagged servers
agsync sync --namespace coding-agents  # only coding-agents-tagged + untagged servers
agsync sync                            # all servers (no filter)
```

Rules:

- `namespaces` is optional. MCPs without it (or with an empty array) are **always included** — treat them as global.
- A skill that references an MCP excluded by the active namespace produces a warning, not an error.
- Scaffold a tagged MCP with `agsync mcp add  --namespace ci-cd` (repeat the flag to add multiple).

## License

MIT

## Source & license

This open-source MCP server is cataloged on AgentStack and links to its original source — we do not rehost the code.

- **Author:** [yiftahb](https://github.com/yiftahb)
- **Source:** [yiftahb/agsync](https://github.com/yiftahb/agsync)
- **License:** MIT

Install and usage instructions live in the source repository linked above.

## Pricing

- **Free** — Free

## Security capabilities

Automated source analysis of v0.1.0 — what this tool can access:

- **Network access:** no
- **Filesystem access:** no
- **Shell / process execution:** no
- **Environment & secrets:** no
- **Dynamic code execution:** no

*"Yes" means the capability is present in the source — more access means more to trust, not that it is unsafe.*


## Versions

- **0.1.0** — security scan: passed — Imported from the upstream source.

## Links

- Listing page: https://agentstack.voostack.com/l/mcp-yiftahb-agsync
- Seller: https://agentstack.voostack.com/s/yiftahb
- Browse the marketplace: https://agentstack.voostack.com/browse

---
Listed on AgentStack — the marketplace for AI agent skills and MCP servers. Every listing is security-reviewed. Creators keep 70%.