Post by Sahaj Software

As AI agents become more capable, they also become more connected to files, databases, APIs, and even shell commands. The layer making these interactions possible is the Model Context Protocol (MCP). But how secure are the MCP servers that power enterprise AI workflows? We are excited to share that Akash Sathish from Sahaj Software will be speaking at The Fifth Elephant's 'Enterprise AI in Production', where he'll explore the security realities of today's rapidly growing MCP ecosystem. In his session, "When the Agent Workflow Survives Production but the MCP Server Ships an RCE," Akash will draw on findings from an audit of 50+ production MCP servers to discuss: 🔹 How to evaluate third-party MCP servers before integrating them into agent workflows 🔹 Common vulnerability patterns, including command injection, SSRF, path traversal, hardcoded credentials, and tool poisoning 🔹 Findings from real-world vulnerabilities and how they map to public CVEs 🔹 Using MCPeek, an open-source static analyzer, to identify security issues before deployment 🔹 The role of static analysis in securing AI infrastructure and where additional runtime protections are needed 📍 PhonePe, Bengaluru 📅 19th June, 2026 ⏰ 1:45 PM - 6:00 PM (IST) 🔗 https://lnkd.in/gyxiuZJp Hasgeek | Sahaj Software #TheFifthElephant #EnterpriseAI #AgenticAI #MCP #AIEngineering #ApplicationSecurity #DevSecOps #SahajSoftware