Avneesh Kasture is an Associate Solutions Engineer at Gruve whose work focuses on secure AI integrations. His projects include mcploitable, deliberately vulnerable tool servers for studying agent-security failures and their controls.[1]
Start with Mcploitable – The "Metasploitable" of the Model Context Protocol →
Education
Bachelor of Technology in Computer Science and Engineering
Completed 2024
[2]Links
Personal portfolio
GitHub · agileAlligator Associate Solutions Engineer, Professional Services · Gruve
Career
Kasture’s resume records a Deloitte digital-forensics internship from July 2024 to February 2025, then a Gruve security-operations role from April 2025 to April 2026. His Associate Solutions Engineer title began in April 2026; the resume separately says he assumed its responsibilities in December 2025. His integration work includes secured MCP scaffolds with endpoint allowlists, identity checks and audit logging.[2][1]
Contributions
mcploitable supplies ordinary-looking mail, analytics and other tool servers with intentional vulnerabilities. Learners test how malicious input can redirect goals, misuse tools or cross permissions, then compare controls at several strengths. Some scenarios are interactive boxes and others guided simulations. The project’s tests distinguish model-dependent guardrails from deterministic checks such as allowed endpoints and identity verification, making the enforcement mechanism visible.[3]
Tools, writing & talks
Writing, research & talks
Start here
An isolated learning environment with deliberately vulnerable MCP servers and progressively stronger controls.
Read the scenario and control ladder, then use only the bundled isolated environment with its fake data. Docker and basic agent-security knowledge help; these intentionally vulnerable servers must stay separate from real credentials, systems and untrusted networks.
Creator and repository owner[2][3]Original source