Thumbnail
landing page
dashboard

About SentinelMesh

Inspiration

AI agents are becoming increasingly powerful, capable of interacting with tools, accessing files, and performing tasks automatically. However, these capabilities also introduce security risks when an agent makes an unauthorized tool call or attempts to access sensitive information. We were inspired to build SentinelMesh, an AI Agent Security Firewall designed to make AI-powered automation safer, more controlled, and more trustworthy.

What It Does

SentinelMesh acts as a security layer between AI agents and the tools they use. It evaluates tool calls against security policies, checks permissions, blocks unauthorized or potentially dangerous actions, and records security decisions in audit logs. Its dashboard helps developers monitor security events, review alerts, manage policies, and test attack scenarios.

How We Built It

We approached SentinelMesh as a combination of a security gateway and a monitoring dashboard. Our planned architecture separates the frontend interface from the backend security engine, with persistent storage for policies, events, and agent information. We focus on server-side validation, least-privilege access, and a fail-closed approach so that uncertain or unauthorized requests are rejected rather than automatically allowed. Safe mock tools can demonstrate security decisions without executing dangerous operations.

Challenges We Faced

One of our main challenges was understanding how to connect the frontend, backend, and database while ensuring that security decisions happen on the server rather than relying on the user interface. We also had to consider how to distinguish legitimate tool calls from suspicious ones, prevent unauthorized access, and maintain useful audit records. Balancing strong security with a simple, intuitive developer experience was another important challenge.

What We Learned

Building SentinelMesh helped us explore backend development, API design, access control, security policy enforcement, database integration, and Git-based collaboration. We also learned that protecting AI agents requires more than monitoring their behavior: every protected tool action must pass through an enforced security gateway.

Our Vision

Our vision is to make AI agents safer to deploy and easier to trust. SentinelMesh aims to give developers greater visibility and control over agent actions while providing a foundation for more secure AI automation.

SentinelMesh — Secure Every AI Action.

Built With

Submitted to

Build, Ship, Shape: Amazon Developer Hackathon

Created by

As the Front-End Developer of SentinelMesh, I contributed to designing and developing a user-friendly, interactive, and responsive interface using HTML, CSS, and JavaScript. I focused on presenting security alerts, threat analysis, and application information clearly, ensuring a smooth user experience. I also worked on improving the visual design, layout, and usability of the application to make the security features easy to understand and access.

vidyaaditya-cg Aditya
I contributed to the development of SentinelMesh, an AI Agent Security Firewall designed to make AI agent interactions with tools safer and more transparent. My work involved using AI Studio for AI-assisted development, building and refining the security monitoring dashboard, presenting security events and alerts, and demonstrating simulated security scenarios. I also worked on project documentation and GitHub preparation to support open-source collaboration. The project explores how policy-based access control and MCP-based tool integration can help developers identify and prevent unauthorized AI agent actions.

Suraj Kumar Sahoo
Navya Sathya Sri Arakala
shashi143-star