How to Use the Cerbos MCP in Google ADK

Q: How does Cerbos work with Google ADK?

You add the Cerbos server as an McpToolset when creating your LlmAgent. The ADK then makes the Cerbos tools available to your Gemini model, so it can make authorization calls as part of its reasoning process.

Q: Can I use Cerbos to control access to my data in BigQuery with a Google ADK agent?

Yes, that's a primary use case. Use the planresources tool in your agent to get a query plan from Cerbos. Then, use that plan to build a secure SQL query that only returns data the user is authorized to see.

Q: Can I limit which Cerbos tools my agent can see?

Yes. The McpToolset in the Google ADK has a toolnames filter. You can use it to expose only specific tools, like checkresources, to a particular agent.

Gate access to Google Cloud resources using Cerbos and the Google ADK for your Gemini agents.

See Vinkius in Action

Works with every AI agent you already use

…and any MCP-compatible client

MCP Servers - Free for Subscribers

Connect Cerbos MCP to Google ADK

Create your Vinkius account to connect Cerbos to Google ADK and route execution through our secure gateway. The platform manages server hosting, runtime updates, and security layers. Configuration requires no manual server provisioning.

GDPR Free for Subscribers

Setup Cerbos with Google ADK

Ask AI about this MCP

ChatGPT

Claude

Perplexity

Generate Access Plans for BigQuery

The `plan_resources` tool is what you need to secure data access. It gives your agent a query plan that defines which resources a user is allowed to see, based on your central Cerbos policies. For the Google ADK, this means your agent can construct a safe BigQuery `WHERE` clause. It asks Cerbos 'what can this user see?' and gets back the exact conditions to build a secure query, preventing data leaks.

Run Centralized Permission Checks

Use the `check_resources` tool to ask a simple question: 'Can this principal do this action on these resources?' Cerbos evaluates your policies and returns a direct answer. This is critical for enterprise agents built on the Google ADK. You can check permissions on a Vertex AI endpoint or a Cloud Storage bucket before your agent attempts to use it, all against one source of truth.

Connect Your Gemini Agent to a Cerbos MCP Server

The `get_authzen_config` and `get_server_info` tools help your agent understand its environment. It can find the right AuthZEN API endpoints or check the server version for diagnostics. Your Gemini agent can use its long-context window to reason about these configurations. An agent can self-diagnose connection issues or adapt its behavior based on the Cerbos instance it's connected to. This MCP server makes your agent more aware.

Setup guide

Set up Cerbos MCP in Google ADK

Prerequisites

Python 3.10+ installed
google-adk package (pip install google-adk)
Active Vinkius subscription with a valid endpoint token

1

Install Google ADK

Run pip install google-adk to install the Agent Development Kit. MCP support is included via the McpToolset class.
2

Connect via SSE transport

Use McpToolset.from_server() with SseServerParams pointing to your Vinkius endpoint. Replace [YOUR_TOKEN_HERE] with your token from cloud.vinkius.com.
3

Create an LlmAgent

Pass the returned mcp_tools list directly to LlmAgent(tools=mcp_tools). The ADK maps each MCP tool to a native Gemini function call — no manual schema definitions required.
4

Run with any Gemini model

The agent works with any Gemini model (gemini-2.0-flash, gemini-2.5-pro, etc.). Copy the full example on the right to get started with Cerbos tools in your ADK agent.

agent.py

from google.adk.agents import LlmAgent
from google.adk.tools.mcp_tool.mcp_toolset import McpToolset
from google.adk.tools.mcp_tool.mcp_session_manager import SseServerParams

# Connect to the MCP via SSE
mcp_tools, exit_stack = await McpToolset.from_server(
    connection_params=SseServerParams(
        url="https://edge.vinkius.com/[YOUR_TOKEN_HERE]/mcp"
    )
)

# Create your agent with auto-discovered tools
agent = LlmAgent(
    name="Cerbos_agent",
    model="gemini-2.0-flash",
    instruction="You have access to Cerbos tools via MCP.",
    tools=mcp_tools,
)

Get your connection token →

Independent Platform Disclaimer: Vinkius is an independent platform and is not affiliated with, endorsed by, sponsored by, verified by, or otherwise authorized by Cerbos. All third-party trademarks, logos, and brand names are the property of their respective owners. Their use on this website is strictly for informational purposes to identify service compatibility and interoperability.

Why Choose Vinkius

Vinkius connects your tools to AI with real-time monitoring and automatic cost savings — all from one dashboard.

Connect Cerbos now

Real-time monitoring

Live

visibility into every interaction

Connect your favorite tools to your AI and see exactly what's happening — every request, every response, in real time.

Built-in savings

60%

lower AI costs

Vinkius compresses data between your apps and your AI automatically. Lower bills every month — no configuration required.

Single dashboard

One

place for every integration

Every tool your AI connects to, managed from a single screen. One account, complete control.

Common questions about Cerbos MCP in Google ADK

You add the Cerbos server as an `McpToolset` when creating your `LlmAgent`. The ADK then makes the Cerbos tools available to your Gemini model, so it can make authorization calls as part of its reasoning process.

Yes, that's a primary use case. Use the `plan_resources` tool in your agent to get a query plan from Cerbos. Then, use that plan to build a secure SQL query that only returns data the user is authorized to see.

It's not about being better, it's about decoupling. With Cerbos, you manage access rules separately from your agent code. This lets you update permissions without having to redeploy your agent on Google Cloud.

Yes. The `McpToolset` in the Google ADK has a `tool_names` filter. You can use it to expose only specific tools, like `check_resources`, to a particular agent.

Only the data needed for the check, like a principal's identifier, the resource they're trying to access, and the action. This information is processed ephemerally and is never logged or stored by the Vinkius-managed server.

Use it with your favorite AI tools

Connect this server to Cursor, Claude, VS Code, and more.

OpenAI Agents SDK sdk-python

Google ADK sdk-python

Pydantic AI sdk-python

Vercel AI SDK sdk-typescript