Asking a person

Three places in this server do something a mailbox does not undo. All three ask a person first.

Not a confirm: true argument the model can set. Not a token the model reads out of its own previous result. A dialog, raised through MCP elicitation, that goes to the client and is shown to whoever is sitting there.

The specification says a client should keep a human in the loop:

there SHOULD always be a human in the loop with the ability to deny tool invocations

This server does not rely on that. It raises the question itself, and until an answer comes back, nothing happens.

What asks, and when

Tool	When it asks
delete_messages	always
move_messages	always, for both move and copy
manage_mailbox with action: "delete"	always
manage_mailbox with action: "rename"	two-call token only — a rename is reversible
manage_mailbox with action: "create"	never
set_message_flags · save_draft	never

move_messages was on the token alone until now, on the grounds that a move destroys nothing. Its own comment already said what is wrong with that: destination is a free-form mailbox name, so on a shared account or a public namespace one call hands every named message to everyone who can read that folder. Disclosure is the part that cannot be taken back, copying is the mode that does it while leaving no trace in the source folder, and a token only proves the model agreed with itself.

The approval covers the exact UID set and both mailboxes: one for [1] will not execute [1, 2] where the model picked the second list, and one for Archive will not execute against Public/Shared. move and copy get different keys, because they have different consequences.

set_message_flags is deliberately unguarded — flags come back off — and it refuses \Deleted outright, because on a server with autoexpunge that would be a deletion wearing a flag's clothes.

What the dialog contains

Counts, UIDs and folder names. Never a subject or a body: every message in the mailbox was written by somebody who could send you mail, and the prompt is read by a model at the exact moment it is deciding.

This will move 12 message(s) between mailboxes.

The messages keep their content but get new UIDs, so the current ones stop
working.

Values below are supplied by the caller, not by this server:
  From: INBOX
  To: Archive

The approval is bound to its target, so one obtained for a call cannot be replayed against another. For a set of targets the binding is a fingerprint of the exact list: an approval for ["a"] does not execute ["a", "b"].

Clients that cannot show a dialog

Not every MCP client implements elicitation, and a stateless gateway may not be able to speak for the one it is currently serving. Rather than refuse to work — which pushes people towards switching the guard off entirely — the tool falls back to a two-call token: the first call returns a random string, the second has to quote it back.

Be clear about what that proves, because this server is:

the token proves the call was made twice with the same arguments, and nothing more.

A model can read the token out of the first result and call again in the same turn without anybody seeing it. It catches a widened target set; it does not catch a model that was talked into the whole thing. The fallback text says so rather than implying somebody approved.

Switching the dialog off

ELICITATION=false

Default is true. false does not remove the guard — it takes the fallback path above, which means the token. There is no setting in which a guarded call goes unannounced.

Use it where a dialog is the wrong shape rather than an unwanted one: a scheduled job, a test harness, a client whose dialog interrupts something else.

It is deliberately not prefixed

ELICITATION has no IMAP_ in front of it, so one export ELICITATION=false — or one -e ELICITATION=false in a compose file — reaches every MCP server in that environment, not just this one. That is the point of it and also its risk.

Two things make it visible rather than silent:

• a server started with it off prints one line at startup, in the log of every server it actually reached:

  imap-mcp: ELICITATION=false — guarded tools fall back to the two-call token

• the fallback text names the server that did not ask, instead of blaming a client that was working fine. :::

Anything other than true or false — 1, off, yes — stops the server with exit code 1 and a message naming both valid values. This is the only variable in this family that defaults to on: a typo that fell back to the default would leave the dialog running while the operator believed it was off, and there would be nothing to tell them.

Annotations are the other half, and they are only a hint

Every tool of this server declares all four MCP tool annotations — readOnlyHint, destructiveHint, idempotentHint, openWorldHint — so a client can tell before it calls what a call would do. See Tools.

They are advice, and the specification says so:

clients MUST consider tool annotations to be untrusted unless they come from trusted servers

An annotation is something a client may ignore. The dialog is not: it is enforced here, on the server side, and no answer means no change. The two are different claims — the annotation says what a call does, the dialog decides whether it happens — which is why a tool can be marked destructive without being guarded. get_attachments shows why the annotations are worth stating rather than inferring: it is the only computed one in the family — readOnlyHint: true until IMAP_DOWNLOAD_DIR is set, because from then on it writes files.

Behind a gateway

Both protocol revisions are handled from one code path. On 2025-11-25 the question is pushed to the client; on 2026-07-28 there is no server→client channel at all, so the call returns input_required, ends, and the client retries carrying the answer.

That answer arrives as ordinary request content, which the SDK does not validate — so the state that ties an answer to its question is sealed (HMAC). A reply whose seal does not open, or opens onto a different target, counts as no answer and produces a fresh question rather than an error. The likeliest cause is not an attack: it is a gateway that put the server to sleep while the person was reading.

If you run this behind mcp-hub, the hub passes elicitation through in both directions; see its elicitation guide.