Connect MultiMail via MCP to Claude Code so your terminal-based AI assistant can send, read, and manage email with full oversight controls.
Claude Code is Anthropic's terminal-native CLI for working with Claude directly in your development environment. By adding MultiMail as an MCP server, Claude Code gains the ability to send emails, manage inboxes, handle contacts, and orchestrate email workflows without leaving your terminal.
Claude Code supports both global and project-scoped MCP configurations. You can add MultiMail globally in ~/.claude.json so it's available in every session, or scope it to a specific project using .mcp.json in your project root. This flexibility lets you use different API keys or mailboxes per project.
MultiMail's oversight modes pair naturally with Claude Code's agentic workflow. Set mailboxes to gated_send during development and testing, then move to monitored or autonomous for production automation.
Sign up at multimail.dev and generate an API key from the dashboard. You'll need this to authenticate the MCP server.
Create a mailbox for Claude Code to use. Set the initial oversight mode to gated_send so all outbound emails require your approval during setup.
For global access, add the MultiMail server to ~/.claude.json. For project-specific access, add it to .mcp.json in your project root.
Start a new Claude Code session and ask Claude to list your mailboxes. If the MCP server is connected correctly, it will return your MultiMail mailboxes.
| Tool | Description | Example |
|---|---|---|
| send_email | Send an email from your agent mailbox to any recipient. | Send a deployment notification to the team after a successful release. |
| reply_email | Reply to a received email, maintaining the thread context. | Reply to a bug report email with the fix details and commit hash. |
| check_inbox | Check a mailbox for new or unread messages. | Check if the CI pipeline sent any failure notification emails. |
| read_email | Read the full content of a specific email including headers and body. | Read a detailed error report email from a monitoring system. |
| create_mailbox | Create a new agent mailbox on your MultiMail account. | Create a project-specific mailbox for automated notifications. |
| list_mailboxes | List all mailboxes on your MultiMail account. | Check which mailboxes are available before sending a message. |
| get_thread | Retrieve an entire email thread to see the full conversation history. | Review the full thread of a client discussion before replying. |
| manage_contacts | Search your contact list by name, email, or tags. | Find the email address of a team member by name. |
| manage_contacts | Add a new contact to your MultiMail contact list. | Save a new collaborator's email after they contribute to the project. |
| tag_email | Tag or categorize an email for organization and filtering. | Tag deployment-related emails with 'devops' for later filtering. |
| list_pending | List all emails currently awaiting human approval. | Review pending outbound emails before batch-approving them. |
| decide_email | Approve or reject a pending email in the oversight queue. | Approve a drafted client email after verifying its content. |
No code, no dashboard. Paste this to your AI agent — it connects MultiMail, creates an inbox, and builds the flow for you.
Place a .mcp.json file in your project root and commit it to version control (with the API key excluded via .env or secrets). This ensures everyone on the team has the same MCP setup when using Claude Code in that project.
Claude Code has full access to your codebase. Ask it to read a file, generate a changelog, then email it to stakeholders. The combination of code context and email tools makes Claude Code powerful for developer communication workflows.
With gated_send mode, you can ask Claude to draft an email, review it in the pending queue, reject it with feedback, and ask for a revised draft. This iterative loop helps you get the tone and content right before anything is sent.
Since Claude Code runs in your terminal, you can pipe command output into your prompts. Run your tests, then ask Claude to email the results to the team. Claude can reference the terminal output when composing the email body.
Email infrastructure built for AI agents. Verifiable identity, graduated oversight, and a hosted MCP server. Formally verified in Lean 4.