PieterPost MCP - Connect your AI agent to postal mail
by•
PieterPost MCP connects AI agents to postal mail. From ChatGPT, Claude, Codex, Claude Code, or any MCP client, agents can prepare letters and postcards, use Mailbook contacts, upload attachments or postcard images, create checkout links, and track orders. It brings PieterPost online mail, API, and payment-link workflows into agent tools.


Replies
Two things for me. First, the fully resolved postal address read back verbatim, plus which Mailbook entry it matched, since 'John in London' quietly resolving to the wrong saved contact is the failure I'd never catch. Second, an idempotency key on the send, so if the agent's tool call times out and retries I get one postcard and not two. Duplicate physical sends are the money version of a double-submit.
Pieter Post
@dipankar_sarkar Really appreciate this. Address readback and making sure retries do not create duplicate mail are exactly the kind of sharp edges we want to make boring. Thanks for trying it and writing this out, we will make this much better soon.
Love seeing MCP tools that result in real physical objects. Good stuff!
Pieter Post
@hsearcy Thanks Houston. Same here, MCP gets a lot more fun when it leaves the screen a little.
Gotta say, the branding here is genuinely charming. That name "Pieter Post" with the classic envelope vibe really sells the whole concept before you even read what it does. Smart move leaning into the postal heritage while modernizing the actual experience.
Pieter Post
@dnd1112360 Thank you, this means a lot. We wanted it to feel like mail, not another cold tool. Pieter Post doing old-school postal work from the browser is basically the whole thing.
Really nice concept, way more convenient than running to the post office. The way it just works behind the scenes and you forget it's even mail feels seamless.
Loved how quick it was to send a letter from my phone, and the tracking kept me from worrying if it actually arrived.
The postal-mail MCP angle is weirdly practical. I’d be curious how you handle confirmation before an agent sends anything physical — approval queue, spend cap, address validation? That boundary feels more important than the API itself.