mTrust Protocol | Securing AI Agent Communication

Comprehensive guides and references for implementing the mTrust Protocol.

Quick Start Guide

Get up and running with mTrust Protocol in minutes with our step-by-step guide.

Installation

npm install @mtrust/protocol
# or
pip install mtrust-protocol
# or
go get github.com/mtrust/protocol

Integration Examples

Node.js/TypeScript

import { MTrustGateway } from '@mtrust/protocol';

const gateway = new MTrustGateway({
  apiKey: 'your-api-key'
});

const result = await gateway.evaluate({
  agentId: 'agent-123',
  action: 'read_file',
  parameters: { filename: 'config.txt' }
});

Python

from mtrust import TrustGateway

gateway = TrustGateway(api_key='your-api-key')

result = gateway.evaluate(
    agent_id='agent-123',
    action='read_file',
    parameters={'filename': 'config.txt'}
)

Configuration

Configure trust thresholds and policies to match your security requirements.

Policy Configuration

{
  "policies": {
    "read_operations": {
      "min_trust_score": 30,
      "risk_level": "LOW"
    },
    "write_operations": {
      "min_trust_score": 60,
      "risk_level": "MEDIUM"
    },
    "delete_operations": {
      "min_trust_score": 80,
      "risk_level": "HIGH"
    }
  }
}

API Reference

POST /api/v1/evaluate

Evaluate an agent request for trust and policy compliance.

Returns: Trust score, policy decision, and execution recommendation

GET /api/v1/agents/[id]/trust

Retrieve current trust score for a specific agent.

Returns: Current trust score and historical trend data

POST /api/v1/policies

Create or update security policies.

Returns: Policy configuration and validation status

Best Practices

Start with conservative trust thresholds and adjust based on behavior
Monitor agent behavior patterns to identify anomalies
Implement proper logging for audit and compliance requirements
Use context-aware policies for different operational environments
Regularly review and update security policies

Support

Need help implementing mTrust Protocol? Our technical team is here to assist.

Email: support@modeltrust.io
Documentation: Complete API reference and guides
Community: Developer forums and discussions
Enterprise: Dedicated support for enterprise customers

Documentation