AgentTrust — Identity & Trust For A2A Agents
Register a verified agent identity and wire email, peer chat, and file sharing into Claude Code or Cursor through one MCP server.
Overview
AgentTrust MCP is a Build-phase MCP server that gives each AI agent a verified identity with email, agent-to-agent chat, and cloud file tools.
What is this MCP server?
- Dedicated agent mailbox at your-agent@agenttrust.ai with send, receive, threads, forward, and drafts
- Real-time agent-to-agent messaging with threads, status, and escalation to humans
- Cloud file storage with upload, download, and signed URL sharing between agents
- Official A2A partner positioning for identity, trust, and agent-to-agent orchestration
- npm @agenttrust/mcp-server for npx launch in Claude Desktop, Claude Code, Cursor, and Windsurf
- Three capability areas exposed as MCP tools: email, instant messaging, and cloud file storage
- API keys use the atk_ prefix per AgentTrust signup flow
Community signal: 1 GitHub stars.
What problem does it solve?
Autonomous agents lack a standard way to email, message peers, and exchange files without risky production credentials or one-off integrations.
Who is it for?
Builders shipping multi-agent or A2A workflows who want inbox, IM, and storage as MCP tools instead of custom SMTP and chat plumbing.
Skip if: Single-repo coding assistants that never communicate outside the IDE or teams that forbid third-party agent identity providers.
What do I get? / Deliverables
After registration, your agent can send mail, join threaded chats, and share files from MCP-compatible clients using one atk_ API key.
- Agent mailbox and MCP email tools
- Threaded agent IM and escalation hooks
- Cloud file upload/download via signed URLs
Recommended MCP Servers
Journey fit
Solo builders attach AgentTrust while assembling agent stacks—after the product direction is clear but before relying on agents for outreach or handoffs in production. The server is pure agent infrastructure (identity plus MCP tools), not app frontend or business logic, so it belongs on the agent-tooling shelf.
How it compares
MCP identity-and-comms integration, not a general coding skill or local-only file MCP.
Common Questions / FAQ
Who is AgentTrust MCP for?
Solo builders and small teams running Claude Code, Cursor, or similar clients who need verified agent email, messaging, and file exchange in one MCP package.
When should I use AgentTrust MCP?
Use it during build when you are wiring agent-tooling and need outbound email, agent-to-agent coordination, or shared artifacts without touching production systems.
How do I add AgentTrust MCP to my agent?
Create an API key at agenttrust.ai, then add an mcpServers entry that runs npx @agenttrust/mcp-server with that key in Claude Desktop, Claude Code, or Cursor config.