Agent Inboxes

Explainer

Mailbox vs webhook for agents

Two different shapes show up when people say “give my agent email.” Mixing them up is the most common reason an integration feels broken.

Persistent mailbox

A mailbox is a store of messages the agent (or its runtime) can read later: threads, folders, IMAP/JMAP/API history, sometimes SMTP send from the same identity. Useful when the agent must wait for a reply, handle OTPs over time, or keep conversation state across runs.

In this directory, that usually maps to agent-inbox and some agent-identity products. Examples currently listed:

Compare agent-inbox options (live-checked).

Inbound webhook / parse

An inbound webhook (or inbound parse worker) delivers a push event when mail arrives — often as JSON to your endpoint — without necessarily giving the agent a long-lived inbox UI or IMAP view. Useful for “on email → run function” automations, ticket creation, or piping mail into object storage.

In this directory, start with inbound-routing. Examples currently listed:

Quick chooser

Choose a mailbox if the agent needs to own an address, read history, reply in-thread, or block until a human responds.

Choose inbound routing / webhooks if you already have an address and only need events or parsed payloads into your app.

Transactional send (transactional) is outbound-focused; pair it with a mailbox or inbound route when you need replies.

Feature tags on listings (e.g. imap, webhooks, mcp) come from our catalog and the latest link check (2026-11-06). They are labels, not lab results — verify against vendor docs.