Secure Azure MCP Server
Details
Now that you are planning or have already deployed your Azure MCP Server, it's time to secure it.
This practical session dives into the security layer of the MCP, focusing on how to protect your agents from metadata poisoning, prompt injection, and unintended tool execution.
From attack scenarios to best practices in infrastructure, we’ll walk you through the steps needed to secure your MCP deployment and ensure your agents are operating in a controlled, safe environment.
This is the second session in the AI Motion track focusing on operational and secure agent infrastructure.
# What You’ll Learn?
- Common attack surfaces in MCP servers
- How implicit trust in tool metadata leads to indirect agent control
- Input validation & description sanitization for tool definitions
- Using schema validation and content filtering for LLM safety
- Hardening your Azure MCP Server
- Logging and monitoring
# Who Should Attend?
Cloud Security Engineers
cybersecurity people
Agent developers using Copilot / LLMs
Anyone running or integrating Azure MCP in production
# Agenda
20:00 – Welcome & Opening
20:05 – Secure Azure MCP Server
20:35 – Questions & Key Takeaways
20:40 – Open Q&A
# Lecturers
Topaz Hurvitz - GenAI & Cloud Security Architect @ Israel Post
Elli Shlomo - Head of Security Research @ Guardz | Community Advocate
# Notes
- Level: 200-300 (Practical, technical, hands-on)
- The event will be recorded
- The event will be delivered in Hebrew
# Community Channels
Secure Azure MCP Server