Skip to main content

Agent Security

Agent Security is a comprehensive permissions layer designed specifically for Model Context Protocol (MCP) servers. It enables organizations to implement fine-grained access control for their AI agents while maintaining security and compliance requirements.

Core Concepts​

Agent Security combines several powerful technologies to deliver a complete permissions solution:

  • Fine-grained ReBAC: Leverages Permit.io, OPA, and OPAL to implement relationship-based access control
  • Multi-stage Security: Implements authentication and authorization that binds user, agent, and MCP server identities
  • Complete Auditing: Provides comprehensive auditing of every agent action
  • Human Oversight: Enables human-in-the-loop (HITL) approvals for critical decisions (coming soon)

Getting Started with Documentation​

Our documentation is organized to help you understand and implement Agent Security effectively. Here's how to get started:

  1. Review the Overview to understand the core concepts and fundamental principles
  2. Study the Architecture to learn about technical architecture and component interactions
  3. Follow the Getting Started Guide for practical setup and usage instructions

Support and Resources​

Contents