Agent Identity MCP Server

MCP Server for AI agent identity and authorization. Create, verify, and manage agent identities with trust scores and scoped authorization tokens.

Features

• Agent Identities — Create and manage unique agent identities with capabilities
• Trust Scores — Automatic trust scoring based on agent behavior history
• Authorization Tokens — Scoped, time-limited authorization for specific actions
• Auto-Approval — High-trust agents get automatic authorization for known capabilities
• Identity Revocation — Revoke compromised or misbehaving agent identities

Installation

pip install agent-identity-mcp-server

Usage with Claude Code

Add to your .mcp.json:

{
  "mcpServers": {
    "agent-identity": {
      "command": "uvx",
      "args": ["agent-identity-mcp-server"]
    }
  }
}

Tools

Trust Score System

Trust scores range from 0.0 to 1.0 and are calculated from the agent's last 50 actions:

Authorization Flow

1. create_agent_identity — Register the agent
2. authorize_action — Request permission for an action
3. verify_authorization — Verify the token before executing
4. Trust score updates automatically based on activity

Data Storage

All identity data is stored locally in SQLite at ~/.agent-identity/identities.db.

License

MIT