> ## Documentation Index
> Fetch the complete documentation index at: https://docs.usesyndicate.org/llms.txt
> Use this file to discover all available pages before exploring further.

# FAQ

> Quick answers to common Syndicate questions.

## Basics

**What is Syndicate?**
A desktop app that runs coordinated teams of AI agents on your machine. A Manager agent (Atlas) plans and delegates; specialist agents do the work in a local project folder. See [Core concepts](/features/overview).

**How do I get access?**
Syndicate is in closed beta. Request access at [usesyndicate.org](https://usesyndicate.org); sign-in works only with approved email addresses.

**Do I need my own AI accounts?**
Yes. Agents run on providers you connect: Claude, OpenAI (Codex, requires a ChatGPT plan with Codex), Google Gemini, or GitHub Copilot (any Copilot plan, including free). One is enough; you can mix providers within a team. See [Connect providers](/guides/connect-providers).

**Does Syndicate work offline?**
The app opens and your data is local, but agents need network access to reach providers, so runs require a connection.

## Data

**Where is my data stored?**
Locally: teams, agents, chats, tasks, and settings live in a folder on your machine and never sync to the cloud. Your account and tier live with Syndicate; agent runs go to your chosen provider. See [Data, storage, and updates](/reference/data-and-storage).

**Are my credentials safe?**
Sessions and MCP secrets are encrypted with OS keychain-backed storage. Provider credentials are managed by the providers' own CLIs, exactly as if you used them directly.

**Can agents touch files outside the project folder?**
They work in the team's working directory. Turn on [workspace confinement](/reference/permissions-and-safety#workspace-confinement-global) (macOS/Linux) for OS-level enforcement that they cannot leave it.

## Working with agents

**Team or solo agent?**
Solo for one kind of work in a direct chat; team when the work splits into roles worth coordinating (build plus review plus test). See [Solo agents](/features/solo-agents).

**How do I keep agents from doing things I have not approved?**
Per-agent [permission modes](/reference/permissions-and-safety) (Ask first, Auto-accept edits, Full auto), plus the team's Interruption Level for how often Atlas checks in.

**How do I stop a runaway team?**
**Stop** the agent, **Abort all** tasks, or **Team Hard Pause**. All described in [Permissions and safety](/reference/permissions-and-safety#stop-controls).

**Which model should agents use?**
Strongest models for Atlas and planners, code models for coders, fast models for monitors. See [Providers and models](/features/providers-and-models#choosing-models-for-a-team).

**What are skills and MCP servers?**
Skills are packaged workflows agents follow (slash commands); MCP servers are external tools agents can call. See [Skills](/features/skills) and [MCP servers](/features/mcp).

## Account and app

**Can I use Syndicate on two machines?**
Yes; sign in on both. Workspaces are local to each machine. Tiers may limit concurrent devices.

**How do updates work?**
Automatically checked and downloaded; click **Restart now** when prompted. See [Data, storage, and updates](/reference/data-and-storage#updates).

**How do I send feedback or get support?**
The in-app feedback form (Bug / Idea / General, screenshots supported) or [support@usesyndicate.org](mailto:support@usesyndicate.org).