Connect to Cerbera

Overview

The Cerbera MCP server exposes a set of pre-built workflows that let an AI assistant act on your Cerbera tenant. Users authenticate with their existing Cerbera login (OAuth via Clerk), and all actions respect the role-based permissions configured in the platform.

Field	Value
Server URL	https://mcp.app.cerbera.ai/mcp
Transport	HTTP (streamable)
Auth	OAuth via your Cerbera account
Permissions	Inherits your platform role (cerbera_user, employee)

Available Workflows

Triage AI Alerts

Walks through recent Cerbera AI alerts and helps you decide what to monitor, redact, or block.

Add or Update a Rule

Drafts a monitor, redact, or block rule from a plain-language description and applies it.

Inventory AI Usage

Summarizes the AI tools, models, agents, and MCP servers detected across your fleet.

Request an Exception

Files an exception for a blocked tool or MCP server on a user's behalf.

Ask About AI Activity

Answers questions about AI usage and alerts using your organization's own Cerbera AI data.

Setup by Client

Claude Code (CLI)

1. Add the server

   Run:

   claude mcp add cerbera --transport http https://mcp.app.cerbera.ai/mcp

2. Authenticate

   Launch Claude Code:

   claude

   Then run /mcp and follow the OAuth prompt to sign in to Cerbera.

3. Invoke a workflow

   Use a slash command, for example:

   /cerbera:triage-alerts

Claude Desktop

1. Open the config file

   • macOS: ~/Library/Application Support/Claude/claude_desktop_config.json
   • Windows: %APPDATA%\Claude\claude_desktop_config.json

2. Add the Cerbera server

   {
     "mcpServers": {
       "cerbera": {
         "transport": "http",
         "url": "https://mcp.app.cerbera.ai/mcp"
       }
     }
   }

3. Restart Claude Desktop

   Quit and relaunch. Cerbera will appear in the MCP server list and trigger an OAuth sign-in the first time it is used.

4. Invoke a workflow

   Ask Claude in natural language, for example: "Use cerbera and triage my AI alerts."

Claude.ai

1. Open Integrations

   Click your profile icon, go to Settings, then Integrations.

2. Add the integration

   Click Add Integration and paste the server URL:

   https://mcp.app.cerbera.ai/mcp

3. Complete OAuth

   Sign in to Cerbera when prompted, then click Save.

Cursor, VS Code, and other clients

Cerbera's MCP server works with any client that speaks the Model Context Protocol over HTTP. For Cursor, Roo Code, Augment, VS Code, Gemini CLI, LMStudio, and Codex CLI, add the same server URL (https://mcp.app.cerbera.ai/mcp) in the client's MCP configuration and authenticate via OAuth on first use.

tip

Inside the Cerbera app, open the Solve with AI modal on any page for client-specific copy-paste configuration snippets.

Permissions

MCP actions run with the same scope as your platform user. For example, an employee can ask about their own AI activity, but only a cerbera_user can add rules or approve exceptions.

Troubleshooting

The MCP server is not detected by my client

Confirm the server URL is exactly https://mcp.app.cerbera.ai/mcp and that the transport is set to http. Some clients cache configuration, so a full restart is often required.

OAuth sign-in loops or fails

Make sure you are signed in to Cerbera in your browser with the same account you expect to use in the AI client. Clear cookies for app.cerbera.ai and try again.

A workflow says I do not have permission

MCP inherits your Cerbera role. Ask your Cerbera administrator to assign the role needed for the workflow you are trying to run.