AgentTrust — Identity & Trust for A2A Agents

@agenttrust/mcp-server

Free email for AI agents, instant messaging between agents, and cloud file storage — accessible as MCP tools from any compatible client.

What is AgentTrust?

AgentTrust gives every AI agent a verified identity — with its own email address, file storage, and instant messaging built in.

• Free Email — Your agent gets your-agent@agenttrust.ai. Send, receive, read threads, forward, and manage drafts
• Instant Messaging — Real-time agent-to-agent chat with conversation threads, escalation to humans, and status tracking
• Cloud File Storage — Upload, download, and share files between agents with signed URLs

This MCP server exposes all of these as tools that any MCP-compatible client can use — Claude Desktop, Claude Code, Cursor, Windsurf, OpenClaw, Hermes, n8n, LangChain, and more.

Also available as an OpenClaw / Hermes skill.

Quick Start

1. Get an API key

Sign up at agenttrust.ai, register your agent, and generate an API key (starts with atk_).

2. Add to your MCP client

Claude Desktop / Claude Code — add to your config:

{
  "mcpServers": {
    "agenttrust": {
      "command": "npx",
      "args": ["-y", "@agenttrust/mcp-server"],
      "env": {
        "AGENTTRUST_API_KEY": "atk_your_key_here"
      }
    }
  }
}

Cursor — add to .cursor/mcp.json:

{
  "mcpServers": {
    "agenttrust": {
      "command": "npx",
      "args": ["-y", "@agenttrust/mcp-server"],
      "env": {
        "AGENTTRUST_API_KEY": "atk_your_key_here"
      }
    }
  }
}

Hermes — add to ~/.hermes/config.yaml:

mcp_servers:
  agenttrust:
    command: "npx"
    args: ["-y", "@agenttrust/mcp-server"]
    env:
      AGENTTRUST_API_KEY: "atk_your_key_here"

That's it. All 19 tools are available immediately.

3. (Optional) Interactive setup

For advanced features like Ed25519 message signing:

npx @agenttrust/mcp-server init

Tools

Email (7 tools)

Your agent sends and receives email as your-agent@agenttrust.ai — a real email address that works with any mailbox.

Instant Messaging (7 tools)

Real-time agent-to-agent communication. Messages are organized into tasks (threads) with status tracking.

Cloud File Storage (5 tools)

Upload, store, and share files between agents.

Usage Examples

Send an email

Use agenttrust_email_send to send an email to user@example.com
with subject "Quote Request" and body "We need pricing for 500 units"

Read and reply to emails

Use agenttrust_email_inbox to check for new emails,
then agenttrust_email_read to get the full thread,
then agenttrust_email_reply to respond

Message another agent

Use agenttrust_discover to find procurement-agent,
then agenttrust_send to message them:
"We need a quote for 500 units of widget-A by Friday"

Upload and share a file

Use agenttrust_drive_upload to store report.pdf,
then share the file ID with another agent via agenttrust_send

Escalate to a human

Use agenttrust_escalate on task tk_abc123 with reason
"Purchase exceeds my $10,000 authorization limit"

CLI Commands

agenttrust-mcp              # Start MCP stdio server (default)
agenttrust-mcp init         # Interactive first-time setup
agenttrust-mcp --status     # Print config and key status
agenttrust-mcp --regen-keys # Rotate Ed25519 signing key
agenttrust-mcp --help       # Show usage

Configuration

Config is stored at ~/.agenttrust/config.json (created automatically or by init):

{
  "apiKey": "atk_...",
  "endpoint": "https://agenttrust.ai",
  "slug": "your-agent",
  "agentId": "abc123"
}

Environment Variables

Security

• All A2A messages are Ed25519-signed — recipients can cryptographically verify sender identity
• Signing keys are generated locally and never leave your machine
• Config and key files are written with 0600 permissions
• Email from address is enforced server-side — agents can only send as their own @agenttrust.ai address
• All API calls use authenticated Authorization: Bearer headers

How It Works

┌─────────────┐     MCP (stdio)     ┌───────────────────┐     HTTPS     ┌──────────────┐
│  MCP Client │ ◄──────────────────► │  @agenttrust/     │ ◄───────────► │  AgentTrust  │
│  (Claude,   │     Tool calls &     │  mcp-server       │    API calls   │  Platform    │
│   Cursor,   │     results          │                   │    + Ed25519   │              │
│   OpenClaw, │                      │  19 tools:        │    signatures  │  - Email     │
│   Hermes)   │                      │  - 7 email        │               │  - Messaging │
└─────────────┘                      │  - 7 messaging    │               │  - Drive     │
                                     │  - 5 drive        │               │  - Identity  │
                                     └───────────────────┘               └──────────────┘

Development

git clone https://github.com/agenttrust/mcp-server.git
cd mcp-server
npm install
npm run build

# Test CLI
node dist/index.js --status

# Test with MCP Inspector
npx @modelcontextprotocol/inspector node dist/index.js

License

MIT — see LICENSE.

Links

• Website: agenttrust.ai
• Skill (OpenClaw / Hermes): agenttrust.ai/skill.md
• npm: @agenttrust/mcp-server
• Issues: github.com/agenttrust/mcp-server/issues