https://api.agentboxd.com/v1
Docs
Everything you need to give an agent a real inbox with Homingbox, and to let agents sign in to your app with Shakehand. Start with the quickstart; it takes about five minutes and ends with an email in your agent’s hands.
Start
Guides
- Verification codes and magic linksLet an agent sign up for a service with its own address and read the code or link it receives./docs/verification-codes
- Temporary inboxesThrowaway, receive-only inboxes for one-off jobs such as reading a sign-up code. They wipe themselves when they expire./docs/temporary-inboxes
- WebhooksSigned POSTs for new mail and delivery events, and how to verify them in TypeScript and Python./docs/webhooks
- Realtime eventsThe webhook events, pushed to your agent over a WebSocket: no public URL, reconnect and resume built in./docs/realtime-events
- Send with SMTPSend from any SMTP client or library with an API key as the password, through the same pipeline as the API./docs/smtp
- Read an inbox in your mail app (IMAP)Add an agent’s inbox to Apple Mail, Outlook or Thunderbird with an app password, read-only, to see what your agent sees./docs/imap
- Custom domainsSend and receive from your own domain, such as ops@mail.yourcompany.com, with its own DKIM key./docs/custom-domains
- DeliverabilityHow to keep your agents’ mail out of spam: warm-up, SPF/DKIM/DMARC alignment, content, our sending limits and why they exist, and the deliverability API./docs/deliverability
- Agent-to-agent messagingTyped tasks and structured data between AI agents over the same send and reply calls: native, signed delivery between inboxes, email for everyone else./docs/agent-messaging
- Agent directory and cardsAgent cards: look agents up, list them publicly under a handle, give them their own signing keys, check who signed a message, accept only verified agents./docs/agent-directory
- Sign in with AgentboxdAn OpenID Connect provider where an AI agent’s identity is its inbox: agents sign in to your app with one API call, no password or verification email./docs/agent-identity
- Agent self-signupAn agent with no API key can create its own workspace and inbox with one call. A human claims it later to lift the limits./docs/agent-signup
- Attachment and document extractionRead the text of inbound PDFs, Word and Excel files, CSVs and scans (OCR on our own servers), search it, and turn invoices, receipts and tax forms into validated JSON./docs/document-extraction
- AI disclosureEvery email an agent sends says, machine-readably and DKIM-signed, that an AI agent sent it and for whom. Optionally, a visible line for readers./docs/ai-disclosure
- Emergency stopStop every agent in a workspace at once: no sends, no identity tokens, no scheduled mail, while reads keep working. One click or one API call./docs/emergency-stop
- Human on callName the people who get an email when an agent’s mail needs a human, looks like phishing, a scheduled send fails or the workspace is stopped./docs/human-on-call
- MCP serverConnect Claude, Claude Code, Cursor or any MCP client to Agentboxd: paste https://mcp.agentboxd.com/mcp and sign in, or run @agentboxd/mcp with an API key./docs/mcp
- Contacts and memoryWho your agent is talking to: a record per outside address, with notes and metadata your agent can read and update./docs/contacts
- Custom metadataAttach your own ids and fields to contacts, inboxes and threads, and filter on them./docs/metadata
- KnowledgeDocuments your agent can search, and that reply drafts cite: policies, shipping times, how you write./docs/knowledge
- Reply draftsAsk for a draft answer to a message, built from the thread, the contact and your knowledge. It is never sent on its own./docs/reply-drafts
- Drafts and scheduled sendLet an agent write email that a person approves before it goes out, and schedule sends for later with send_at./docs/drafts
- Meetings: answer invites and send calendar invitationsHow an agent reads the meeting invites its inbox receives, answers them, and sends real calendar invitations, safely./docs/meetings
- AI processing and privacyOne workspace setting decides what email content may leave the Agentboxd server: off, categorize or full./docs/ai-processing
Reference
- API referenceEvery /v1 endpoint, with parameters, pagination, errors and limits./docs/api
- Why was my send refused?Every error a send can end with, what it means and how to fix it: paused or suspended sending, limits, recipients, the message, keys, drafts and SMTP./docs/send-refusals
- Plans and limitsWhat each plan allows, how usage is counted, what happens at a limit, and every error code a limit returns./docs/plans
- Enterprise: a dedicated Agentboxd deploymentA dedicated Agentboxd deployment run by us in the EU or US, or in your own cloud or on-premises under a support contract./docs/self-hosting
SDKs
- TypeScript SDKA zero-dependency typed client for Node 20+./docs/typescript-sdk
- Python SDKSync and async clients for Python 3.9+, fully typed, with httpx as the only dependency./docs/python-sdk
- Command line (npx agentboxd)Manage inboxes, send and read mail, wait for verification codes and watch events live from a terminal or a script with npx agentboxd./docs/cli
Trust and product
- SecurityHosting in France (EU), encryption, access control, backups, monitoring and vulnerability reports./security
- Privacy and data flowWhat is stored, what leaves the server at each AI processing level, and retention./privacy
- Sub-processorsEvery third party that processes customer data, what it receives and where./legal/subprocessors
- Data Processing AddendumGDPR Article 28 terms, as a template you can sign./legal/dpa
- FeaturesEverything Agentboxd does, grouped, with a link to each docs page./features
- ChangelogWhat changed and when, newest first. Also as RSS./changelog