MCP Server Profiles
MCP Server Profiles let administrators control exactly what an AI assistant or agent can see and do when it connects to Liferay through the Model Context Protocol. Each profile defines a specific set of actions (tools) the AI is allowed to use, grouped by application area for easy selection. Administrators can further fine-tune each tool by hiding specific data fields from its responses, so sensitive information never reaches the AI even when the tool itself is enabled. Profiles also apply configurable data masks — such as automatically redacting emails, phone numbers, or ID numbers — adding another layer of protection across any data the AI can access. This gives organizations granular, auditable control over AI integrations, letting them safely expose Liferay functionality to AI agents without compromising data privacy or security.
Key Benefits:
- Granular access control: define exactly which actions an AI agent can perform in Liferay, scoped per use case (e.g., content editing vs. user administration).
- Automatic data masking: redact emails, phone numbers, national IDs, and other sensitive values before they ever reach the AI.
- Multiple isolated profiles: run different AI integrations side by side, each with its own permissions and data exposure rules.
- Built-in safe defaults: pre-configured system profiles for common scenarios, protected from accidental deletion or modification.
- Auditability and compliance: changes to sensitive configurations (like removing a data mask) require a documented reason, supporting governance requirements.
- Reduced integration risk: lets organizations adopt AI/LLM tooling with confidence that data privacy and security boundaries are enforced by design.
Status: Release Flag (opt-in)