Deploy a Secure MCP Server on Cloud Run
A workshop and codelab walking developers through the concepts of agentic AI, the Model Context Protocol (MCP), and how to deploy a secure MCP server on Cloud Run: making it easy to connect AI agents to external tools and databases at scale.
Events
| Name | Organizer | Date | Location | Attendees | Links |
|---|---|---|---|---|---|
| DevFest Bacolod 2025 | GDG Bacolod | 2025-11-22 | University of St. La Salle, Bacolod City, Philippines | 109 | 📊 Slide Deck |
QR Code for this session
Abstract
As AI agents become a core part of how modern software operates, the challenge shifts from building a single capable model to connecting many agents reliably to the external world: databases, APIs, and services. The Model Context Protocol (MCP) addresses this directly: it standardizes how AI agents discover and call tools, reducing the boilerplate of tool-calling code and improving the reliability and reusability of agent integrations.
This workshop starts with the fundamentals of agentic AI: what agents are, how tools and function-calling work, and how the ecosystem has evolved from simple LLM prompts to multi-agent systems. It then introduces MCP: what it is, why it matters, and how Google's open-source MCP Toolbox for Databases fits into the picture. The second half of the workshop dives into Cloud Run as the ideal serverless runtime for hosting MCP servers: covering its architecture, benefits for AI workloads, and design patterns for GenAI apps and agents. The workshop concludes with a live codelab where attendees deploy their own secure MCP server on Cloud Run.
Agenda
- Agentic AI fundamentals: agents, tools, and how function calling works
- The integration problem: reliably connecting agents to databases, APIs, and services
- What is MCP? Protocol overview and why standardization matters
- Google's MCP Toolbox for Databases: simplifying agent-to-database connections
- Cloud Run as the ideal serverless runtime for MCP servers
- Design patterns for GenAI apps and multi-agent systems
- Codelab: Deploying your own secure MCP server on Cloud Run
Key Takeaways
- MCP standardizes tool discovery and calling - eliminating custom glue code across every agent project
- A well-designed MCP server is reusable across multiple agents and use cases simultaneously
- Cloud Run's serverless model is a natural fit for MCP servers: scalable, secure, and pay-per-use
- Google's MCP Toolbox removes the hardest part of connecting agents to production databases
- Security is built into Cloud Run by design - auth, HTTPS, and IAM are handled for you