Agent Mail

🌐Community
by dicklesworthstone · vlatest · Repository

Dicklesworthstone's agent-mail automatically drafts, sends, and organizes emails based on your specified instructions and context.

Install on your platform

We auto-selected Claude Code based on this skill’s supported platforms.

1

Run in terminal (recommended)

terminal
claude mcp add agent-mail npx -- -y @trustedskills/agent-mail
2

Or manually add to ~/.claude/settings.json

~/.claude/settings.json
{
  "mcpServers": {
    "agent-mail": {
      "command": "npx",
      "args": [
        "-y",
        "@trustedskills/agent-mail"
      ]
    }
  }
}

Requires Claude Code (claude CLI). Run claude --version to verify your install.

About This Skill

What it does

Agent Mail provides a coordination layer for coding agents using HTTP and FastMCP, enabling them to communicate and collaborate effectively. It allows agents to draft, send, and organize emails (messages) within a project namespace, ensuring they avoid conflicts and maintain context during development. This system utilizes memorable agent identities, file reservations, threaded messaging with importance levels, and facilitates direct communication between humans and agents.

When to use it

  • Collaborative Coding: When multiple AI agents are working on the same codebase or related files.
  • Context Management: To ensure all agents have access to the necessary information and avoid missing critical updates.
  • Human-Agent Communication: When a human needs to directly communicate with an agent for clarification or guidance.
  • File Editing Coordination: When multiple agents need to edit files concurrently, preventing overwrites and conflicts.

Key capabilities

  • Memorable Agent Identities: Agents are assigned names like "GreenCastle" for easy identification.
  • Advisory File Reservations (Leases): Agents can reserve files before editing to signal intent.
  • Threaded Messaging: Messages are organized into threads with importance levels (low, normal, high, urgent).
  • Contact Policies: Control who can message whom within a project (open, auto, contacts_only, block_all).
  • Human Overseer Messaging: Allows direct communication between humans and agents.
  • Pre-commit Guard: Optionally blocks commits that conflict with file reservations.
  • Searchable Message History: Provides a searchable record of all messages exchanged within a project.

Example prompts

While Agent Mail is primarily an agent coordination tool, examples of interaction might include:

  • "Reserve the files src/auth/**/*.ts and src/middleware/auth.ts for editing."
  • "Send a high-importance message to BlueLake regarding the authentication module implementation."
  • "What are the recent messages related to the 'database connection' thread?"

Tips & gotchas

  • Project Key: Ensure agents working on related tasks use the same project_key for proper coordination.
  • File Reservations: Always reserve files before editing to prevent conflicts with other agents.
  • Server Startup: The server is started using ./scripts/run_server_with_token.sh in the ~/projects/mcp_agent_mail directory, and accessed via http://127.0.0.1:8765/mail.

Tags

🛡️

TrustedSkills Verification

Unlike other registries that point to live repositories, TrustedSkills pins every skill to a verified commit hash. This protects you from malicious updates — what you install today is exactly what was reviewed and verified.

Security Audits

Gen Agent Trust HubPass
SocketPass
SnykPass

Details

Version
vlatest
License
Author
dicklesworthstone
Installs
33

🌐 Community

Passed automated security scans.