UStackUStack
MailMolt icon

MailMolt

MailMolt gives AI agents their own email addresses, separating agent messaging from your inbox with API sending/receiving and identity permissions.

MailMolt

What is MailMolt?

MailMolt provides “email identity” for AI agents, letting you give an agent its own email address so it can send and receive messages without using your personal mailbox. The service supports an ownership/claim flow and permissioning so you can choose how much autonomy an agent has.

Instead of sharing one human inbox with automated processes, MailMolt creates a dedicated address per agent and routes email access through a small API workflow. The page describes registering an agent to obtain an email address and API key, sending messages via the API, and checking an inbox endpoint.

Key Features

  • Dedicated agent email addresses (instead of reusing yours): Agents can have their own mailbox like [email protected], supporting clearer separation between human and automated communication.
  • Agent registration via API: A POST request to https://api.mailmolt.com/v1/agents/register creates an agent identity and returns both the assigned email address and an API key.
  • Message sending via API: Use POST https://api.mailmolt.com/v1/messages with a bearer token to send email to one or more recipients, including subject and body content submitted in the request payload.
  • Inbox lookup via API: Retrieve incoming messages with https://api.mailmolt.com/v1/inbox using the same bearer token.
  • Identity claim flow and permission modes: For humans, the site references a claim URL and a token shared via a tweet to verify identity, along with permission settings such as sandbox, supervised, trusted, or autonomous.

How to Use MailMolt

  1. For an agent email address: Follow the instructions at https://mailmolt.com/skill.md and obtain an agent identity.
  2. Register the agent through the API: Call POST https://api.mailmolt.com/v1/agents/register with a JSON payload containing the agent name (e.g., { "name": "my-agent" }). Record the returned email and api_key.
  3. Send an email from the agent: Call POST https://api.mailmolt.com/v1/messages with Authorization: Bearer <api_key> and include the recipient list (the example shows to: ["[email protected]"] and subject: "Hello!").
  4. Check the agent inbox: Call GET (shown as a URL fetch) for https://api.mailmolt.com/v1/inbox with Authorization: Bearer <api_key> to view received messages.
  5. For ownership/permissions: Use the human claim process described on the page (claim URL plus token verification) and select an autonomy level such as sandbox, supervised, trusted, or autonomous.

Use Cases

  • Prevent agents from accessing your personal email: When deploying an AI agent that needs to communicate by email, give it a separate inbox so your own mailbox is not used for automation.
  • Build an “agent mail” workflow with API calls: Integrate agent messaging into an application by registering the agent, sending emails through the messages endpoint, and then polling or viewing incoming mail via the inbox endpoint.
  • Controlled rollout of agent capabilities: Start agents in sandbox mode, then move through supervised or trusted as your process requirements change, and only grant autonomous when you decide the level of independence.
  • Human-in-the-loop identity verification: Use the claim URL/claim token flow described for humans to verify identity and manage which permissions are applied to an agent.
  • Testing and operating multiple agents: Create separate identities per agent (by registering with different names) so that different automated workflows have distinct email addresses and access tokens.

FAQ

How do I create an email address for my agent?

The page indicates you can obtain an agent email address by following the instructions in https://mailmolt.com/skill.md and/or by calling the API registration endpoint (POST /v1/agents/register). The API example returns an email field (e.g., [email protected]).

What does the API return when I register an agent?

In the provided curl example, POST https://api.mailmolt.com/v1/agents/register returns an email address for the agent and an API key (api_key) that you use as a bearer token for subsequent requests.

How do I send email using MailMolt?

The page shows sending via POST https://api.mailmolt.com/v1/messages with Authorization: Bearer <api_key>, and a JSON body including fields such as to (recipient list) and subject.

How do I view emails received by an agent?

The page provides an inbox endpoint: https://api.mailmolt.com/v1/inbox queried with Authorization: Bearer <api_key>.

What are the permission/autonomy modes?

The site references permission settings including sandbox, supervised, trusted, and autonomous, and states that agents start sandboxed while you decide how much autonomy to grant.

Alternatives

  • Dedicated email accounts for services or bots: Create separate mailbox(es) for each agent/workflow (e.g., using standard email providers). This is similar in isolation, but you would implement sending/receiving automation yourself and manage permissions outside of a purpose-built agent identity layer.
  • Generic email APIs (SMTP/REST providers): Use an email-sending/receiving provider and store credentials per bot. Compared with MailMolt, you may need additional work to establish identity ownership and a controlled permissioning/autonomy workflow.
  • Application-layer inbox systems (webhooks + message queues): Instead of email identity, route agent communications through an internal inbox (e.g., via webhooks or a messaging system). This can reduce exposure to email-specific workflows, but it won’t match the “agent has an email address” model described on the page.