mcp-server

Aegis: Secure AI Agents

Name: Aegis: Secure AI Agents
Author: Justin0504

Get runtime policy enforcement for AI agents with Aegis. For founders using AI agents.

364 stars37 forksTypeScriptQuality 8/10Updated 7/12/2026100% free · open source

What it does

Aegis provides runtime policy enforcement for AI agents, allowing founders to have control over AI decision-making with features like cryptographic audit trails and human-in-the-loop approvals

Install / run

git clone https://github.com/Justin0504/Aegis.git

When to use it

•When deploying AI agents in critical environments where accountability is key
•When needing to ensure compliance with regulations that require explainability and transparency in AI decision-making
•When wanting to implement a kill switch for AI agents to mitigate potential risks or errors

Quick start

1Navigate to the cloned repository with 'cd Aegis'
2Review the policy configuration in the README to understand how to set up cryptographic audit trails and human-in-the-loop approvals
3Implement the policy enforcement mechanism in your AI agent workflow, referring to the provided TypeScript examples
4Test the kill switch functionality to ensure you can interrupt AI agent operations if necessary
5Explore the GitHub repository for more detailed documentation and example use cases

Ready-to-paste prompt

To test the Aegis kill switch, you could use a command like 'node aegis.js --kill-switch' assuming you've set up your AI agent integration as described in the README

Heads up: First-time users need to ensure their development environment supports TypeScript, as Aegis is built with TypeScript, and may need to install or update their TypeScript compiler to match the version required by Aegis

Saves to your device