ChatGPT
Runs through OpenAI's Codex CLI. Good for reasoning, writing, summaries, and code.
Connect and share your favorite AI models.
A Worker is an AI tool on your computer that a Space is allowed to use. You install the tool and sign in with your own account; Syntila Workspace runs it when someone in the Space asks.
Claude
Runs through Anthropic's Claude Code. Good for long documents, planning, code changes, and review.
Copilot
Repository-aware help and terminal tasks through GitHub Copilot's command-line agent.
Gemini
Runs through Google's Antigravity CLI. Good for fast research, large inputs, and a second point of view.
Grok
Reasoning and coding tasks through xAI's Grok Build command-line agent.
MCP
Connect a local model, such as one served by Ollama, or your own agent as a Worker over the Model Context Protocol (MCP).
Check your plan before you share a Worker. Your provider’s terms determine whether others can use a Worker you connect. Check your plan before sharing; personal subscriptions may be limited to one person. Read more in the FAQ.
Give Workers extra tools with MCP servers.
Model Context Protocol (MCP) servers add tools, such as a database or an issue tracker, that a Worker can call. You register them on your own computer and approve what they offer before any Worker can use them.
Register
Add local or remote servers Use syntila mcp add for a server on your computer or syntila mcp add-remote for one reached over HTTPS.
Approve
Review each tool once Syntila records exactly what each tool says it does. If a server later changes a tool, calls are blocked until you review it again.
Keep it local
Separate settings per run Each run gets its own temporary MCP settings. Your normal CLI configuration is not changed, and server tokens stay on your computer.
Run
Under your account Approved servers run as you, on your computer, with output size limits and cleanup when the run ends.