# Quick Start (/docs)



ccsync keeps user-level AI agent setup aligned across machines. It scans supported agent harnesses, syncs custom skills and MCP server definitions through your account, then applies the latest manifest on each device.

## Quick Start [#quick-start]

Install the Homebrew tap and CLI:

```bash
brew tap opencoredev/cc-sync https://github.com/opencoredev/cc-sync
brew install --HEAD opencoredev/cc-sync/ccsync
```

Open the account page and sign in with GitHub:

```txt
https://cc-sync.dev
```

Create a CLI token, then connect the machine:

```bash
ccsync init
ccsync daemon start
```

The CLI has the hosted Convex backend and web account URL baked in, so the normal setup only needs your token and a device label.

## What Happens [#what-happens]

1. `ccsync init` asks for the CLI token and saves config at `~/.config/ccsync/config.json` with mode `0600`.
2. The raw CLI token is hashed locally before anything is saved.
3. The first local manifest is pushed to your account.
4. `ccsync daemon start` watches supported agent config paths and syncs future changes.

## Next [#next]

* <a href="/docs/install">Install</a>: Homebrew install details and upgrade notes.
* <a href="/docs/account">Account Token</a>: Sign in, create a token, and connect a device.
* <a href="/docs/sync-model">Sync Model</a>: What ccsync syncs and what it ignores.
* <a href="/docs/cli">CLI Reference</a>: Commands for status, push, pull, daemon, and TUI.
* <a href="/docs/safety">Safety</a>: Path limits, token handling, and MCP secret behavior.

Project-level config is intentionally ignored. ccsync only touches known user-level/global agent setup.