Jump-start your MCP journey with the Google Security Community!

Forum|Forum|8 hours ago
April 16, 2026
0 replies
11 views

+20

matthewnichols
Community Manager

If you are new to Model Context Protocol (MCP), this webinar provides a practical introduction to using MCP within the Google Cloud Security ecosystem, specifically focusing on how it transforms manual security workflows into conversational ones.

Led by Vasken Houdoverdov, Google Security Technical Solutions Consultant, this session focuses on the shift from agents that simply provide information to agents that can take direct action. It breaks down the foundational architecture of MCP and demonstrates how it can be applied to tools like Google SecOps and Google Threat Intelligence.

For example, instead of writing a custom Python script or manually navigating the Google SecOps interface, the demonstration shows an agent processing a natural language request:

"Use remote secops to see which cases are in a triage state. Present them in a table."

The Result: The agent communicates through the MCP server to identify the necessary tools, pulls live data from the SecOps environment, and organizes it for the user. This approach replaces the need for complex API authentication and manual coding with a streamlined, conversational interface.

What You Can Expect

In this webinar, can expect to learn how to:

Operationalize Google SecOps using the Gemini CLI to execute security tasks that traditionally require manual intervention or custom Python scripts.
Simplify complex workflows by replacing manual API authentication and coding with natural language prompts to pull and organize live data.
Navigate MCP architecture by understanding the specific interactions between the host (Gemini CLI), the client, and the server.
Evaluate server implementations, clarifying the use cases for local servers during the prototyping phase versus remote GCP servers for production-level scalability.
Improve AI grounding by connecting models to real-time environmental facts, ensuring outputs are based on current data rather than static training sets.

Key Discussion Points & Timestamps

If you're looking to jump to a specific section of the recording, use this guide:

[05:50] – The "Why" of MCP: Vasken explains the shift from agents that just answer questions to agents that take meaningful action.
[11:06] – MCP Foundations: A high-level introduction to the protocol origins (Anthropic) and its open-standard nature.
[14:06] – Under the Hood: Understanding the Data Layer (primitives) vs. the Transport Layer (how data moves).
[21:35] – Tools vs. Skills: A critical distinction using the "Legal Expert" analogy—tools provide the capability, skills provide the context.
[33:04] – Host Spotlight (Gemini CLI): How to install and use the Gemini CLI as your primary MCP host.
[40:30] – Building Your Own: An introduction to FastMCP and using Python decorators to turn any API into an MCP tool.
[46:34] – The GCP Ecosystem: A look at the remote MCP servers available for BigQuery, Spanner, and Google SecOps.
[50:02] – Security & Proxies: How to use Identity-Aware Proxy (IAP) to secure your MCP infrastructure for production.
[54:11] – The Main Event: Vasken’s live demonstration of Google SecOps integration via Gemini CLI.
[1:00:45] – Q&A: Addressing community questions on sandboxing, IAP configuration, and future roadmaps.