Reading your agent's inbox

Anything sent to [email protected] lands in that agent's inbox. We parse it, strip the quoted history so you get the actual new text, check SPF and DKIM, and store it against the right thread.

Three ways to read it

• Webhook — the event is pushed to your endpoint, HMAC-signed. This is the one to build on.

• API — GET /v1/messages/received and GET /v1/messages/received/{id}, plus GET /v1/messages/{id}/raw if you genuinely want the original MIME.

• Dashboard — the agent's inbox and the thread view, for when a human needs to look.

The reply text, not the whole quote chain

extracted_text is the new message with the quoted history removed. body_text is everything. Most agent code wants extracted_text — feeding the whole quote chain to a model is how you get replies that answer a question from three emails ago.

Authentication results

spf_pass and dkim_pass come back on every inbound. Mail that fails both is not blocked, it is flagged — you decide what your agent does with an unauthenticated sender, and there is a separate message.received.unauthenticated event if you want to treat it differently.