# clai MCP server

> Organize and control teams of AI agents within sandboxed workspaces using MCP-native tools and scheduled runs.

- Repository: https://github.com/clairun/clai
- Page: https://hostedmcp.org/servers/clairun/clai
- Transport: —
- Stars: 9
- License: MIT

## FAQ

### How does CLAI handle security and sandboxing?

CLAI uses platform-specific sandboxing: bubblewrap on Linux, Seatbelt on macOS, and filesystem grants on Windows to control shell and network access.

### Can I use my own LLM providers?

Yes, you can connect OpenAI-compatible or Anthropic-compatible providers, including local endpoints like Ollama or vLLM.

### Can agents run automatically on a timer?

Yes, you can configure workspaces to run on a schedule using periodic runs.

### Does it support MCP servers?

Yes, you can configure MCP servers via HTTP or stdio transports and attach them to specific agents or workspaces.