MCPmailer vs Mailgun

Mailgun has the most flexible inbound routing of the classic senders. MCPmailer gives the agent the mailbox, the thread, and the tools.

Transactional email APIs · Mailgun claims checked July 2026

What Mailgun is

Mailgun is the developer-oriented sender with genuinely powerful inbound Routes: match on recipient or headers with expressions, then forward, store, or POST. If you want to build email plumbing, it hands you the best raw material of the classic providers.

Where Mailgun is the better choice

If your product is email-heavy and inbound routing logic is your own domain logic, Mailgun is the right foundation. Its Routes can express things our fixed mailbox model simply cannot, and its validation and deliverability tooling are mature.

Where MCPmailer differs

Control is the trade. With Mailgun you write the routing rules, the parser, the thread matcher, the store, the search index, and the agent adapter, then keep all six current. We ship one opinionated version of that stack with MCP on top, so time to an agent holding a real conversation is minutes.

01

A mailbox is not a route

A Route is a rule that fires. A mailbox is a thing that persists: it has unread state, archived state, a thread list, a filter mode, and a history. Agents reason much better about a place than about an event, because a place can be re-read when the model needs to check what it already said.

02

The store and the index come with it

Mailgun can store inbound messages for a retention window, and you still build the searchable model your agent queries. search_inbox is Postgres full-text over the whole workspace, and get_thread returns the exchange in order, with raw MIME kept so a thread can always be rebuilt.

03

Refusals the agent can act on

Our pre-send pipeline can refuse with daily_send_quota_exhausted and a reset time, or because the address is suppressed, or because a velocity tripwire fired. Those are designed to be read by a model. A 4xx from a generic API is not.

Side by side

Every row is a claim we could defend with their documentation open beside it. Where they are ahead, the row says so.

Built for
MCPmailer Agents holding conversations. The mailbox is the product, not an endpoint bolted onto a sender.
Mailgun Developers building email plumbing.
Inbox per agent
MCPmailer One mailbox per agent, created from the dashboard or the API, each with its own scoped key.
Mailgun No. Routes and mailing domains.
Receiving and threading
MCPmailer Received, parsed, threaded, and stored, so a reply arrives as part of a conversation rather than as a webhook body.
Mailgun Yes, Routes with expression matching, forwarding, and storage.
MCP server
MCPmailer Native. Streamable HTTP with a bearer key, or OAuth 2.1 for clients that cannot hold one.
Mailgun No first-party MCP surface.
Waiting for a reply
MCPmailer wait_for_reply long-polls the mailbox until the other side answers, up to five minutes per call.
Mailgun No.
Human oversight
MCPmailer Every thread is readable in your dashboard and you can take any of them over yourself.
Mailgun Logs and analytics.
Sending limits
MCPmailer Reply-first. Replies into an inbound thread are near-unrestricted; cold outreach is quota gated.
Mailgun Plan limits plus compliance review.
Your domain
MCPmailer Instant you.mcpmailer.email subdomain, then verify agents.yourcompany.com with a guided DNS wizard.
Mailgun Your own verified domain.
Search
MCPmailer Full-text search across the workspace, exposed to the agent as search_inbox.
Mailgun Log search, retention by plan.
Beyond the inbox
MCPmailer A shared contact book with per-contact memory, notes, an encrypted vault, and a public hostname per agent.
Mailgun Address validation, deliverability tooling, and inbound storage.
Free tier
MCPmailer 3,000 emails a month, 100 a day, 3 inboxes, no card.
Mailgun Limited free tier around 100 emails a day.
Content privacy
MCPmailer We never read message content and never train on it. Enforcement is behavioral and metadata only.
Mailgun Stores content when you use its storage routes.

Switching from Mailgun Routes

If your Routes work, keep them. The case for moving is everything downstream of the route.

  1. 01

    Move agent addresses to an agent mailbox

    Point them here instead of at a Route, and delete the matching expression.

  2. 02

    Retire the parser and thread matcher

    MIME parsing and In-Reply-To linkage happen on ingest.

  3. 03

    Keep Mailgun for bulk and validation

    Address validation and high-volume product sending are not things we do.

Questions

I already have Mailgun Routes working. Why switch?

For the agent addresses only, and only for what sits after the route: thread state, searchable history, per-agent keys, reply-first quotas, and a dashboard a non-engineer can read. Keep Routes for everything else.

Is Mailgun cheaper?

At high volume, yes, and its basic plan is around $15 for 10,000 emails. We include receiving, threading, search, and a dashboard in that price, which is the comparison worth running.

Can I keep using Mailgun to send?

For your product mail, yes, and many customers do. Agent mail should leave from the agent's own subdomain so the two reputations stay separate.

Do you support inbound routing rules?

Per-identity mail rules with whitelist or blacklist mode, matching an exact address or a bare domain. Less expressive than Routes, deliberately.

Looking for a Mailgun alternative?

Keep your transactional sender for the mail your product sends. This is for the addresses your agents converse from, which is a different job with different failure modes.

The free tier is 3,000 emails a month across three agent inboxes, no card, with receiving, threading, and search included. Enough to watch a real conversation happen before you decide.

More Transactional email APIs