Integrations

Integrations let a tool-capable model work with services and resources beyond the conversation. LLMster sends the selected integration IDs to LM Studio, where the MCP servers are configured.

Set up the host

1. Configure the MCP server in LM Studio’s mcp.json, following that integration’s installation and authentication instructions.
2. Enable Require Authentication and Allow calling servers from mcp.json in LM Studio’s Server Settings. Use an API token with the required permissions.
3. Confirm the integration works on the host and note its exact ID, such as mcp/playwright.

See LM Studio’s MCP API guide and server settings. Configuring an ID in the client does not install the corresponding server or authenticate it to another service.

Add it to LLMster

Open Settings → Integrations. The app includes a starting list of integration IDs. Use Edit to remove entries you do not need, or add the exact ID expected by your server.

Defaults on this screen select integrations for newly created chats. Configure an existing conversation separately.

Enable tools in a conversation

1. Choose a model with Tools capability.
2. Open Chat Settings → Integrations and enable the integration needed for this task.
3. Send a specific request that calls for its tools.
4. Enable Render Tool Calls in Chat Settings → Models if you want the tool activity visible in the conversation.

Changing these toggles affects future requests. A capable model decides whether to call a tool; switching one on does not force every response to use it.

Tools can access external resources

An integration may read or change files, contact an online service, or perform actions with its configured permissions. Enable only the integrations appropriate for the conversation. Hiding tool-call rendering does not revoke access.

If nothing happens, check the exact ID, host configuration, token permissions, per-chat toggle, and model capabilities. Continue with integration troubleshooting.