Connect to Cerbera
Overview
The other side of MCP for Cerbera is letting your own AI assistant act on your Cerbera AI data: triaging an alert, drafting a rule, or asking about AI activity, from inside Claude, Cursor, Claude Code, or another MCP-compatible client, instead of opening the Cerbera app.
This is on the roadmap rather than available today. MCP Governance — discovering and controlling the MCP servers your employees connect to — is available now; connecting your own assistant to Cerbera AI over MCP is not yet shipped.
What This Will Look Like
The intent is a small set of pre-built workflows an assistant can run on your behalf, for example:
- Triaging recent AI security alerts and deciding what to allow, monitor, redact, or block.
- Drafting a rule from a plain-language description.
- Summarizing AI and MCP usage across the fleet.
- Answering questions about AI activity using your organization's own Cerbera AI data.
Actions would respect the same role-based permissions configured in the platform as the rest of Cerbera AI, so an employee could ask about their own activity while only a cerbera_user could add rules.
In the Meantime
Everything a future MCP integration would read is already available through the API and OpenTelemetry export, and everything it would change (rules, policies) can be managed directly in the app today. If connecting your own assistant to Cerbera AI is on your critical path, let your Cerbera security engineer know so it can be prioritized.