Tool governance

​

How tool governance works

1. Register MCP servers in the Manager UI or YAML config with transport settings and connection details
2. Attach servers to agents through per-agent allowlists, so each agent only sees the tools you assign
3. The engine discovers tools from each configured MCP server at startup
4. Agents invoke tools during conversations as needed, with results passed back for response generation

​

Per-agent tool allowlists

• Give a customer support agent access to a CRM tool but not a database query tool
• Let a research agent fetch web content but block file system access
• Share the same MCP server across multiple agents or restrict it to one

​

Transport types

​

Configuration example

• Manager UI
• YAML config

1

Browse the transport catalog

Navigate to MCP Servers in the sidebar. The catalog shows transport type cards: Streamable HTTP, SSE, WebSocket, and STDIO.

2

Add an MCP server

Click + on your transport type and configure the command/URL and connection details. The page includes a Quick Start Guide and a “Discover tools” button on existing server configs.

3

Attach to an agent

Attach the server to agents from the agent detail page.

config.yaml

Copy

Ask AI

mcp_servers:
  - name: fetch
    transport: stdio
    command: docker
    args: ["run", "-i", "--rm", "mcp/fetch"]

  - name: filesystem
    transport: stdio
    command: npx
    args: ["-y", "@modelcontextprotocol/server-filesystem", "/data"]
    env:
      ALLOWED_DIRECTORY: /data

  - name: custom-api
    transport: streamable_http
    url: https://mcp.example.com/api
    headers:
      Authorization: "Bearer ${MCP_API_TOKEN}"

​

Integration approaches

​

Framework integration

from idun_agent_engine.mcp import get_adk_tools

idun_tools = get_adk_tools()