As AI agents increasingly access Miro boards through the Miro MCP server, organizations with sensitive content face a new governance gap: existing Enterprise Guard guardrails don't apply to Miro MCP connections, meaning classified board content can flow to external AI systems without restriction.
The new Block MCP Access guardrail closes this gap by extending Intelligent Guardrails to cover Miro MCP, giving Sensitive Content Admins the same classification-based controls they already use for sharing, export, and AI features. With this upcoming feature in the Enterprise Guard add-on, Sensitive Content Admins can:
- Control Miro MCP access by classification level: configure the Block MCP Access guardrail for any combination of your organization's classification labels (for example, block Miro MCP reads on Confidential and Restricted boards while allowing it on General boards).
- Block both read and write operations: when triggered, access via the Miro MCP server won't be available for AI agents and other programmatic clients. This applies to all Miro MCP tool calls to or on a Miro board, including reads, writes, and updates.
- Configure within the existing Intelligent Guardrails workflow: set up in the same place as your other guardrails, with no separate setup required.
- Enforce consistently across Miro MCP clients: controls are applied at the authorization layer, regardless of which Miro MCP client or AI agent is making the request.