What does connecting Cursor actually do?
Connect your existing agent to the same private workspace you use in the Montage app. The remote server handles research and workspace tools; your client handles its own conversation and authentication controls. Start with the setup specific to your client, then prove access with a real, read-only tool response.
{
"mcpServers": {
"montage": {
"url": "https://app.montages.co/mcp"
}
}
}Add Montage to Cursor’s MCP configuration
Use Cursor’s remote MCP server controls or its .cursor/mcp.json configuration. The snippet uses Cursor’s mcpServers object and the remote URL; it does not install a local Montage executable. Use the scope that matches your project or user setup, then check that Cursor recognizes the server.
Authorize from Cursor’s connection prompt
Use the authentication control Cursor exposes for the server. Complete Montage sign-in and consent in the browser, then return to Cursor. Support and exact settings labels vary by Cursor version. If the client cannot complete remote OAuth, update through your normal software process rather than inserting a token into source control.
Check the first tool result
Ask Cursor to run signals_status. Review the workspace, available source coverage and budgets before requesting collection or a saved audience. A green server indicator is useful connection information, but the actual tool response is the stronger check that account authorization completed.
Keep the boundaries clear.
- No API key to copy for normal OAuth-capable clients.
- The agent uses your workspace’s plan, limits and research credit.
- Client support and UI labels vary; do not promise an app-store integration.
Start with a read-only check: ask your agent to report actual source coverage before collecting records or exporting an audience.
Primary references
Documentation reviewed 2026-11-13. Product capabilities, client availability and plans can change. Third-party marks identify their products, not an affiliation.