How do I verify Montage workspace access with Microsoft Copilot Studio?
Microsoft Copilot Studio integrates MCP into agents through its onboarding workflow, with generative orchestration as a prerequisite. This is a managed agent-building surface, not the GitHub Copilot editor extension. For Montage, keep connector configuration, maker permissions, and end-user workspace authorization distinct. A server can publish tools dynamically without granting the person testing the agent permission to execute those tools against a private workspace.
Using Microsoft Copilot Studio, call only Montage signals_status as the first live operation. Report the actual workspace identity and coverage from the tool response. If authentication, transport, or the tool is unavailable, stop and explain the blocker. Do not start collection, recurring tracking, writes, publishing, messaging, or social actions.Confirm the agent’s orchestration mode
Open the intended Copilot Studio agent and verify that generative orchestration is enabled, as required by Microsoft's MCP guide. Check the harness and agent type covered by your current documentation instead of assuming every flow accepts the same integration. Decide who will test the connection and which Montage workspace that person expects to see before selecting a server or configuring authentication.
Use the MCP onboarding wizard
Follow the documented wizard to connect an existing MCP server, supplying https://app.montages.co/mcp as the remote service endpoint. Do not choose the separate path for creating an MCP server: Montage is already hosted. A reusable connector definition and the agent that consumes it are distinct objects. Avoid publishing a connector across tenants merely to test one workspace's read-only status operation.
Review the available authentication setup
Choose only an authentication option the wizard documents and that Montage actually accepts. Managed environment policies may prevent a maker from completing the required consent. If the offered OAuth or credential configuration cannot meet the service's requirements, stop with that limitation instead of treating maker sign-in as authorization. No new credential, registration, or tenant-wide permission is created by this guide.
Keep dynamic tool exposure constrained
Review the tools and resources attached to the test agent before invoking it. Copilot Studio reflects server-side additions and removals, so an old tool selection should not be treated as a permanent safety guarantee. Preserve approval requirements for mutations. A tool appearing in the maker interface only establishes its published schema, not whether the signed-in Montage account may execute it.
Test status in the agent’s conversation
Ask the test agent to call Montage signals_status only and inspect the actual tool result for workspace identity and coverage. Keep this separate from publishing the agent or distributing its connector. A failed call remains a failed connection even if orchestration answers conversationally. Do not permit collection, recurring tracking, or social actions as a fallback for a status tool that could not be authorized.
Keep the boundaries clear.
- Generative orchestration and supported agent/harness configuration are prerequisites; tenant policy may block setup.
- Documented setup, not live interoperability certification. A visible tool catalog is not workspace authorization.
- First live operation: read-only signals_status. No automatic collection, recurring tracking, or social actions; broader actions need separate explicit authorization.
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 2026-11-15. Product capabilities, client availability and plans can change. Third-party marks identify their products, not an affiliation.