AgentMail MCP. Give your AI agent a true email inbox.
Gemini Works with every AI agent you already use
…and any MCP-compatible client
Just plug in your AI agents and start using Vinkius.
AgentMail gives your AI agent a dedicated email inbox. It lets your agent read, send, and reply to messages like a human would.
You can manage custom inboxes, scan full conversation threads, send new emails, and even process attachments automatically. It makes your AI client a programmable communication layer for email.
What your AI agents can do
Create inbox
Creates a new email inbox for the agent, optionally linking it to a custom domain.
Delete inbox
Deletes a specific inbox by ID; this action permanently removes all emails in it.
Forward message
Forwards an existing email message, allowing you to add custom text to the forwarded copy.
Create, list, or delete specific email inboxes for your agent using tools like create_inbox, list_inboxes, and delete_inbox.
Retrieve full message history for a thread using get_thread, or list all ongoing conversations with list_threads.
Send brand new emails via send_message, reply directly to a conversation thread using reply_to_message, or forward existing messages with forward_message.
Download and read files attached to incoming messages using the get_attachment tool.
Mark an existing message as read or update its metadata using the update_message tool.
Ask AI about this MCP
Supported MCP Clients
Gemini Waiting for input…
019d7549create inbox
Creates a new email inbox for the agent, optionally linking it to a custom domain.
019d7549delete inbox
Deletes a specific inbox by ID; this action permanently removes all emails in it.
019d7549forward message
Forwards an existing email message, allowing you to add custom text to the forwarded copy.
019d7549get attachment
Downloads or reads a specific file attachment from an incoming email message.
019d7549get inbox
Gets details for a specific inbox ID, confirming its existence and status.
019d7549get thread
Reads every message inside a specific conversation thread, requiring a thread ID.
019d7549list inboxes
Returns an array of all inboxes assigned to the API key, including their IDs and email addresses.
019d7549list threads
Lists active conversation threads within a specified inbox, showing subjects and recent message previews.
019d7549reply to message
Sends a reply to an existing email message or thread, which preserves the original conversation history.
019d7549send message
Sends a brand new email message, requiring the sender inbox ID.
019d7549update message
Updates metadata for an existing message, such as marking it as read.
Choose How to Get Started
Build a custom MCP for your own tools, or connect a ready-made integration from our catalog.
Build Your Own
Turn any API into an MCP. Import a spec, define Agent Skills, or deploy with MCPFusion.
- Import from OpenAPI, Swagger, or YAML specs
- Create Agent Skills with progressive disclosure
- Deploy to edge with MCPFusion framework
- Built in DLP, auth, and compliance on every call
- Real time usage dashboard and cost metering
- Publish to catalog or keep private
Make Your AI Do More
Start with AgentMail, then connect any of our 4,700+ other servers whenever your AI needs more. One click, no limits.
- Use this MCP plus 4,700+ others, all in one place
- Add new capabilities to your AI anytime you want
- Every connection is secured and compliant automatically
- Track usage and costs across all your servers
- Works with Claude, ChatGPT, Cursor, and more
- New servers added to the catalog every week
What you can do with this MCP connector
AgentMail gives your AI agent its own dedicated email inbox. It lets your agent read, send, and reply to messages just like a human does. You can manage custom inboxes, scan full conversation threads, send new emails, and process attachments automatically. It turns your AI client into a programmable communication layer for email. Manage Inboxes with create_inbox to set up a new agent email, or list_inboxes to see all the ones you've got; remember you can also use delete_inbox to wipe out a specific inbox by ID, and get_inbox checks the status of any inbox you need to confirm exists.
To Read and Track Conversations, you can run list_threads to get a list of active conversations in an inbox, showing you the subjects and the latest message preview. For the full history, get_thread reads every message inside a specific conversation thread, needing only a thread ID. To Send and Reply to Emails, you can fire off a brand new message using send_message, which requires the sender inbox ID.
If you need to reply to an existing chat, reply_to_message sends a response that keeps the original conversation history intact. You can also forward_message an existing email, letting you type in custom text to go along with the forwarded copy. To Handle Attachments, you just use get_attachment to download or read any file attached to an incoming email message.
You can also Update Message Status by using update_message, which lets you change metadata on an existing message, like marking it as read. You can also get_attachment to download or read files attached to incoming emails. For all these tools, your AI client makes the whole process run.**
How AgentMail MCP Works
- 1 Subscribe to the AgentMail server and enter your API Key.
- 2 Your AI client calls
list_inboxesto see what inboxes are available. - 3 The agent then uses specific tools (e.g.,
send_messageorget_thread) to perform the required email action.
The bottom line is, your AI agent acts like a user with a real email account, letting it send and receive mail through defined tools.
Who Is AgentMail MCP For?
The customer support manager who needs agents to read incoming tickets and draft thoughtful replies. The sales operations specialist who needs automated outreach and response monitoring. Any developer building an autonomous agent that requires reliable external communication.
Deploys agents that read incoming ticket emails via get_thread and draft replies using reply_to_message before human review.
Builds agents that handle initial outreach using send_message and monitor response status by listing threads via list_threads.
Integrates the agent into existing systems, managing inbox state and file handling using get_attachment and update_message.
What Changes When You Connect
- See your agent manage its own mail. Instead of relying on complex webhooks or external APIs, use
send_messageandreply_to_messageto let the AI client handle all communication. - Process files automatically. When a client sends an email with a PDF, the agent can use
get_attachmentto download and process the file content without manual steps. - Keep history clean. Use
get_threadto pull every message in a conversation. The agent reads the full context, not just the last reply. - Track all conversations. Running
list_threadsshows the agent every active, unread conversation, so nothing falls through the cracks. - Manage the infrastructure. Use
list_inboxesto check which inboxes are available, andcreate_inboxif the agent needs a new address. - Maintain state. After the agent reads a message, use
update_messageto mark it as read, keeping the inbox accurate for the next run.
Real-World Use Cases
Responding to a client inquiry
A sales rep gets an email with a pricing question. The agent uses get_thread to read the full context, drafts a thank-you reply using reply_to_message, and sends it back to the client. The process is contained entirely within the agent's action.
Handling batch support tickets
The support team needs to process 50 new tickets. The agent first runs list_inboxes to confirm the ticket inbox is active. It then uses list_threads to identify all unread conversations, allowing the agent to process them sequentially and respond.
Automated follow-up campaign
A marketing manager wants the agent to follow up with leads. They use send_message to dispatch a personalized follow-up email. Later, the agent monitors replies using list_threads to track responses and trigger the next action.
Data extraction from incoming mail
A developer needs the agent to handle incoming reports. When a report arrives, the agent uses get_attachment to download the file, processes the data, and then sends a confirmation summary via send_message.
The Tradeoffs
Using the wrong tool for context
Asking the agent to reply using only send_message. This sends a generic email and loses the context of the ongoing conversation thread.
→
You must use reply_to_message. This tool ensures the agent replies directly to the correct thread, keeping the full conversation history visible to the recipient.
Ignoring message state
The agent reads the email but never tells the system it's done. The inbox remains cluttered with unread flags, giving a false operational picture.
→
After the agent processes a message, call update_message to mark the message as read. This keeps the inbox accurate and prevents reprocessing.
Over-relying on manual copy/paste
A user manually reads the thread, copies the content, and pastes it into a new prompt for the AI. This is slow and risks losing formatting or attachments.
→
Let the agent run get_thread. This pulls the entire, accurate conversation history into the prompt context automatically. The agent works with the whole story.
When It Fits, When It Doesn't
Use this if your goal is to make your AI client act like a human email user. Specifically, if you need the agent to manage conversations, send outreach, or process attached files, this is the tool. Don't use it if you just need to store or retrieve simple text notes, because then a dedicated database or key-value store is better. If your need is to run a complex multi-step data pipeline (e.g., 'download X, check Y, then write Z'), you might need a workflow orchestration tool like LangChain or LlamaIndex to manage the state between calls, but AgentMail handles the messaging state.
Independent Platform Disclaimer: Vinkius is an independent platform and is not affiliated with, endorsed by, sponsored by, verified by, or otherwise authorized by AgentMail. All third-party trademarks, logos, and brand names are the property of their respective owners. Their use on this website is strictly for informational purposes to identify service compatibility and interoperability.
VINKIUS INFRASTRUCTURE
Cloud Hosted
Managed infra
V8 Isolated
Sandboxed per request
Zero-Trust Proxy
No stored credentials
DLP Enforced
Policy on every call
GDPR Compliant
EU data residency
Token Compression
~60% cost reduction
Works with Claude, ChatGPT, Cursor, and more
The Model Context Protocol standardizes how applications expose capabilities to LLMs. Instead of operating in isolation, your AI gains direct access to external platforms, live data, and real-world actions through secure, standardized connections.
This server provides 11 capabilities that interface natively with Claude, ChatGPT, Cursor, and any MCP client. No middleware. No custom integration required.
Available Capabilities
The old way of handling email for agents involves painful copy-pasting and multiple system checks.
Today, if you want your agent to read an email, you often have to build a web hook that triggers a script. That script then has to hit a database, query the message, and then pass the raw JSON data to the AI. If you want to reply, you copy the subject, open a separate draft window, and manually paste the content and recipient. It's brittle, and it always breaks on attachments.
With AgentMail, the agent handles the whole lifecycle. It uses `get_thread` to pull the full history, processes attachments with `get_attachment`, and executes replies directly with `reply_to_message`. It's one simple, reliable action, and you get the full context every time.
AgentMail MCP Server: Send and manage messages with `send_message`.
Before AgentMail, sending an email from an automation system meant hitting an SMTP endpoint, which is often firewalled or requires complex credential management. You'd have to build custom logic for every single sending scenario.
Now, you just let your agent call `send_message`. It manages the sending process, the required inbox ID, and the structure of the email, giving you reliable, programmatic mail flow straight from your AI client.
Common Questions About AgentMail MCP
How do I start using AgentMail MCP Server to create an inbox? +
You start by calling create_inbox. This function lets you establish a new, dedicated email address for your agent. After creation, you can use list_inboxes to verify the new inbox ID.
What is the difference between `list_threads` and `get_thread`? +
list_threads gives you a high-level summary of all active conversations in an inbox. You must then use get_thread and provide a specific thread ID to read the full message history.
Can I send an email without knowing the sender's inbox ID? +
No. The send_message tool requires the sender's inbox_id. You must first use list_inboxes or get_inbox to retrieve the necessary ID before sending.
How does AgentMail handle attachments? +
You use the get_attachment tool. This downloads the file content from a specific message ID, making the attachment data available for your agent to process.
Is `reply_to_message` the same as `send_message`? +
No. reply_to_message is designed to reply contextually to an existing conversation thread, which preserves history. send_message creates a brand new email that doesn't link to a prior conversation.
What should I do if I run into an error when using `list_inboxes`? +
If list_inboxes returns an empty array, it means the API key is active but has not been assigned any inboxes yet. You'll need to use create_inbox first to set up a working email address for your agent.
How does `get_attachment` handle different file formats? +
The get_attachment tool reads the attachment data, which is typically encoded in base64. Your agent client must be prepared to parse this encoding to access the original file content, regardless of the file type.
Can I update an email using the `update_message` tool? +
Yes, update_message allows you to modify the metadata of an existing message. This is useful for actions like marking a thread as read or changing its status, but it doesn't change the message content itself.
Does my AI agent need a preexisting Gmail or Outlook account to use this? +
No. AgentMail is entirely standalone. You can use your agent to programmatically provision brand new, unique email addresses in seconds, completely independent of legacy email providers.
Can my agent download files that users send to its email address? +
Yes. When scanning threads in the inbox, your agent will detect any attached files. It can then use the dedicated attachments tool to retrieve and analyze the contents of the file before drafting a reply.
Is it possible to manage multiple outreach campaigns at the same time? +
Absolutely. Your agent can create and manage multiple inboxes simultaneously. It can list all active boxes, read unread threads for each, and contextualize its replies based on the specific campaign inbox.
Use it with your favorite AI tools
Connect this server to Cursor, Claude, VS Code, and more.
More in this category
Calenso
Manage appointments, customers, and bookings online via AI.
LinkAce
Manage your bookmarks, tags, and collections via the LinkAce REST API.
Grocy (Home ERP)
Automate your household management with Grocy — track inventory, manage shopping lists, and organize chores directly from your AI agent.
You might also like
CatAAS
Access the Cat as a Service (CATAAS) API to fetch random cat images, GIFs, and create custom cat memes with filters and text.
Campaign Monitor
Design branded email campaigns, segment subscriber lists, and measure engagement with professional marketing analytics.
InvestorLift
Automate real estate deal sourcing via InvestorLift — list marketplace deals, manage your buy box criteria, and message sellers directly from any AI agent.