Skip to main content
Glama
resend

Email Sending MCP

by resend
OverviewSchemaSourceRelated ServersDiscussionsAdmin
TypeScript
Local

Create Inbox Draft

create-inbox-draft

Purpose: Create a draft in a Resend inbox: a message saved for later, NOT sent. At least one of to, cc, bcc, subject, text, html, or attachments is required. Nothing leaves the account until send-inbox-draft is called on it.

Instructions

Purpose: Create a draft in a Resend inbox: a message saved for later, NOT sent. At least one of to, cc, bcc, subject, text, html, or attachments is required. Nothing leaves the account until send-inbox-draft is called on it.

NOT for: Sending mail. To send straight away, use send-new-inbox-email to start a new thread from the inbox, reply-to-inbox-thread-email to answer a thread, or forward-inbox-thread-email to pass a message along.

Returns: The draft's id.

When to use: User asks to draft, compose, or prepare a message without sending it, or wants to read the wording back before it goes out. Pass threadId and replyToEmailId together to draft a reply to a message in a thread; omit both to draft a standalone message. A reply draft without "to" is addressed to the author of the message it replies to, at its Reply-To address if it has one, or, for a message the inbox sent, to its to recipients. An unknown threadId, or a replyToEmailId that is not in that thread, returns a 404. A reply draft without a subject keeps an empty one, with no "Re:" prefix, and cannot be sent until you set one. A message can have only one unsent reply draft, so drafting a second reply to it returns the ID of the existing draft. That call ignores every other field except attachments, which replace the existing draft's attachments. The id this returns is the "draftId" that get-inbox-draft, update-inbox-draft, send-inbox-draft, and remove-inbox-draft all take.

Input Schema

TableJSON Schema
NameRequiredDescriptionDefault
ccNoEmail address to copy, or an array of addresses. Across to, cc, and bcc a draft can carry at most 50 recipients in total.
toNoRecipient email address (e.g. "customer@example.com"), or an array of addresses to reach several. Across to, cc, and bcc a draft can carry at most 50 recipients in total.
bccNoEmail address to blind-copy, or an array of addresses. Across to, cc, and bcc a draft can carry at most 50 recipients in total.
htmlNoHTML body of the draft.
textNoPlain text body of the draft.
inboxIdYesInbox ID
subjectNoSubject of the draft. Max 2000 characters. It cannot be only whitespace or contain a line break.
threadIdNoID of the thread this draft replies to. Must be provided together with replyToEmailId. Omit both to draft a standalone message. Get the thread ID from list-inbox-threads.
attachmentsNoFiles to store on the draft. Each one needs filename, base64 content and contentType; a draft cannot attach a file by URL. There is no limit on the number of files, but together they can total at most 25 MB. An empty file or a blocked file type (such as .exe) is rejected. They are sent with the draft when send-inbox-draft is called.
idempotencyKeyNoOptional unique key that prevents duplicate drafts on retries (sent as the Idempotency-Key header). Use the same key when retrying the same logical draft; use a new key for a different draft. Must be 1-256 characters.
replyToEmailIdNoID of the message in that thread this draft replies to, the "Email ID" get-inbox-thread prints for it. Must be provided together with threadId.

Schema Changelog

Changes observed during successful MCP inspections.

  1. Addedv2.26.1

TDQS

A4.8/5.0
Behavior5/5

Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?

Annotations already declare it a non-destructive write with no open-world reach, but the description goes far beyond: nothing leaves until send-inbox-draft is called, unknown threadId or foreign replyToEmailId returns 404, a subject-less reply draft keeps an empty subject and cannot be sent, only one unsent reply draft exists per message, and a repeat call returns the existing ID while ignoring all fields except attachments. That is unusually rich behavioral disclosure.

Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.

Conciseness4/5

Is the description appropriately sized, front-loaded, and free of redundancy?

Bold section headers (Purpose / NOT for / Returns / When to use) front-load the highest-signal content and make the block scannable. It is long, but the length is justified by 11 parameters, an idempotency contract, and multiple 404/edge-case rules, so each paragraph earns its place.

Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.

Completeness5/5

Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?

Even without an output schema, the description states the return value ('The draft's id') and ties it to the downstream draftId consumers. Error modes (404 on bad thread/email pairing), the dedup rule, and the message-until-send semantics are all covered, so an agent has everything needed to call and interpret this tool.

Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.

Parameters4/5

Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?

Schema coverage is 100%, so the baseline is 3, but the description adds real value beyond the schema: it explains the threadId/replyToEmailId pairing contract (both or neither) and the default addressing logic for a reply draft without 'to' (author, then Reply-To, then the inbox's own recipients). It also clarifies that the returned id is the draftId consumed by the four sibling draft tools.

Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.

Purpose5/5

Does the description clearly state what the tool does and how it differs from similar tools?

States a specific verb and resource plus the critical scope distinction: 'Create a draft... a message saved for later, NOT sent.' The 'NOT for' block names the four sibling tools it competes with (send-inbox-draft, send-new-inbox-email, reply-to-inbox-thread-email, forward-inbox-thread-email), so an agent can route correctly without opening schemas.

Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.

Usage Guidelines5/5

Does the description explain when to use this tool, when not to, or what alternatives exist?

Explicit when-to-use ('User asks to draft, compose, or prepare a message without sending it') plus explicit when-not and named alternatives for each send path. It also distinguishes the reply-draft case (pass threadId + replyToEmailId) from the standalone case (omit both), leaving nothing to inference.

Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.

Deploy Server

Other Tools