The Sim MCP server gives an AI app the whole Sim API through the Model Context Protocol. Your agent can list workspaces, run and deploy workflows, query and edit tables, manage files and knowledge bases, read run logs, and more. It covers the same operations as the API and the CLI.
| Deployment | Server URL |
|---|---|
| Sim Cloud | https://mcp.sim.ai/mcp |
| Self-hosted | https://<your-sim-host>/api/mcp, or your SIM_MCP_URL |
The server uses the Streamable HTTP transport. Sign in with OAuth, the default in every app below, or send an API key.
Connect an app
claude mcp add --transport http sim https://mcp.sim.ai/mcpOpen /mcp in Claude Code, select sim, and sign in to Sim in the
browser.
Add https://mcp.sim.ai/mcp as a custom connector under Settings →
Connectors, then connect it and sign in to Sim. For Team or Enterprise,
an owner first adds it under Organization settings → Connectors. See
Claude's custom connector instructions.
codex mcp add sim --url https://mcp.sim.ai/mcpComplete the browser sign-in. To sign in again later, run
codex mcp login sim.
Add sim to mcpServers in ~/.cursor/mcp.json, then enable it in
Cursor and sign in to Sim:
{
"mcpServers": {
"sim": { "url": "https://mcp.sim.ai/mcp" }
}
}Add sim to .vscode/mcp.json, then start it and sign in to Sim:
{
"servers": {
"sim": { "type": "http", "url": "https://mcp.sim.ai/mcp" }
}
}Any other app that supports remote MCP servers with OAuth works the same way: give it the server URL and choose Streamable HTTP if asked.
Claude's hosted connectors call your server from Claude's infrastructure, so a
self-hosted Sim must be reachable from the internet. A localhost URL works
only with apps that run on your machine, such as Claude Code, Codex, Cursor,
and VS Code.
Try it
Ask your app:
- "List my Sim workspaces and the tables in each."
- "Run the
lead-scoringworkflow with this input and show me the result." - "Find failed runs from the last day and explain what went wrong."
- "Create a table of our open support tickets and add these rows."
The agent finds the right operation, reads its inputs, and calls it. See Tools for how that works.
What your agent can do
The server acts as you. It sees the workspaces you can see, with your role in each, and every call is authorized, rate limited, and logged exactly like the same request to the API. Reads leave your resources unchanged, and apps can ask you to confirm each change. See Authentication to limit an app to reads.
Other Sim MCP surfaces
This server is for operating Sim. Two other MCP features do different jobs:
- Search MCP searches your organization's live provider sources with current access checks.
- MCP deployment exposes your own workflows as tools, and MCP tools connect external servers to Sim agents.