Pain: As teams connect more MCP servers, agents drown in tools. They pick the wrong ones, get over-permissioned, and platform engineers have no curated, auditable surface to govern them.

Solution: Logical MCP lets platform engineers build purpose-built MCP servers in Port: cherry-picked tools, workflows-as-tools, and catalog context with scoped permissions.

Value: Agents get the right tools for the job. Platform teams govern AI the same way they govern human self-service, using Port’s existing catalog, ownership, and RBAC.