Start sending

Email your agent can operate.

MCP, CLI, and skills for the coding agents you already use — the full email surface, not a send-only toy. Your agent sends, receives, replies, and asks you before anything blasts.

Get an API keyDocumentation

54 MCP tools · Full CLI · Approve-by-reply on every plan

Halftone illustration: a small robot at a single radar console in an air-traffic control tower for mail, a huge window onto a pale sky where envelope-planes fly dotted routes, one descending to a runway letterbox
#01 — The loop

One loop, closed.

APIYOUYOUR APPDRIN APIMXJSONTHE REPLYYOUR MXPARSED{ }MCPTHREADYOUR AGENTMCPTHE THREAD{ }get_threadcontactsreply_emailREADCHECKDRAFTAPPROVEYOUR INBOXYOU REPLYIT SHIPS
#02 — MCP

The whole platform, as tools.

54 typed tools over the Model Context Protocol — not a send wrapper. Anything you can do from the dashboard, your agent can do from a conversation.

send_emailget_threadreply_emailverify_domaincreate_inboxsimulate_inbound+ 48 more
MCP server docs →
{
  "mcpServers": {
    "drin": {
      "command": "npx",
      "args": ["-y", "@drin00/mcp"],
      "env": { "DRIN_API_KEY": "drin_k_…" }
    }
  }
}

# works with any MCP client — one config block.
DRIN POSTAGE · SERIES 02 — THE TOOL GROUPSSHEET OF 3
02.1Send & reply

send_email, send_batch, reply_email, threads — the conversation surface, both directions.

02.2Operate

Domains, webhooks, API keys, templates, suppressions, metrics — the infrastructure, inspectable and changeable.

02.3Market

Contacts, topics, campaigns — your agent drafts the campaign; approve-by-reply keeps the launch yours.

#03 — Your coding agent

One command, whichever tool.

Claude Code, Cursor, Codex, Lovable, Replit — pick the tab, paste the block, and the same 54 tools appear.

$ claude mcp add drin \
    -e DRIN_API_KEY=drin_k_… \
    -- npx -y @drin00/mcp

# restart the session — 54 tools appear.
> Triage my inbox and draft replies to anything
  from a paying customer.
#04 — CLI

The same surface, from any shell.

A complete CLI that mirrors the API — sends, domains, inbox triage and replies. Scriptable in CI, and just as comfortable in an agent's hands as in yours.

CLI reference →
$ npm i -g @drin00/cli
$ drin login --browser
✓ authenticated

$ drin send --to sarah@gmail.com \
    --subject "Your invoice" --attach ./invoice.pdf
✓ delivered · em_8f2…

$ drin threads list
th_4k2…  sarah@gmail.com  "Re: Your invoice"
#05 — Approve-by-reply

Agents draft. Humans launch. By email.

The blast is gated, not the agent. It can draft a campaign, stage an automation, prepare the send — and the launch waits for a human reply in your ordinary inbox. No dashboard tab open at midnight; the audit trail is your own mail archive.

THE AGENT'S DRAFT
FromDrin approvalsToyou@yourapp.com
Approve: “v2 launch note” → NEWS
Your agent drafted this campaign overnight from the changelog. Preview below. Reply APPROVE to launch, REJECT to stop — either reply settles it.
YOUR REPLY
Fromyou@yourapp.comToDrin approvals
Re: Approve: “v2 launch note”
APPROVE

The same gate guards automation activation. Drafting is free on every plan; nothing reaches a list without the reply.

Pixel-art strip: small robots running the night shift at a tiny post office — sorting, reading and answering mail while a cat sleeps on the counter
#07 — Built to be read by machines

Docs your agent already understands.

/llms.txt/pricing.md/docsAGENT mode in the nav

Positioning, pillars, install commands and pricing — published as plain markdown, so an agent can evaluate Drin without scraping this page. The HUMAN/AGENT toggle up top serves the whole site as text.