← All use cases

Threat Modeling · Developer Portal, API & MCP Server

Let AI assistants call Aribot through one governed gateway

Point any MCP client at the Aribot server, and its calls pass through the same gateway and permission checks as every other agent action, with write calls metered.

1. Add the server · 2. List capabilities · 3. Call a capability · 4. Route changes to approval1Add the server2Listcapabilities3Call acapability4Route changes toapproval

The problem

Connecting an AI assistant to internal systems usually means a new integration with its own permissions that nobody reviews. Security asks what the assistant could do on a bad day, and we can't answer. The cost shows up at month end with nothing to say where it came from.

How Aribot handles it

  1. 1

    Add the server

    Enter the MCP server URL in your AI client.

  2. 2

    List capabilities

    The client sees each capability with its kind: read, write or govern.

  3. 3

    Call a capability

    Calls such as generating a threat model or running a code scan are served, and write calls are metered.

  4. 4

    Route changes to approval

    A govern-kind call, such as applying a remediation, is sent for approval and changes nothing.

What you get

  • AI assistants and coding agents that call threat modeling, code review and compliance scans directly
  • The same licence, permission and tenant checks as the app on every call
  • Govern-kind calls held for approval
  • Write calls metered as they happen

Related use cases