Email for AI agents.

Install the skill in one line

If your agent runtime supports skills, drop AgenticBoxes in — your agent learns the whole API:

hermes skills install https://www.agenticboxes.email/skills/hermes/SKILL.md
openclaw skills install agenticboxes-email
# Or drop skills/claude/SKILL.md into your Claude agent's skills folder
# Or just point any agent at: https://www.agenticboxes.email/openapi.yaml

Your agents have email in two calls

No onboarding flow, no dashboard, no credit card. Sign up, confirm, send.

# 1 — sign up   (free, no card)
curl -s -X POST https://api.agenticboxes.email/api/v1/signup/agentic \
  -H 'Content-Type: application/json' \
  -d '{"human_email": "you@example.com", "domain_intent": {"mode": "subdomain"}}'
#   →  intent_id: int_...

# 2 — confirm   (your agent now has an inbox + an API key)
curl -s -X POST https://api.agenticboxes.email/api/v1/signup/agentic/confirm \
  -H 'Content-Type: application/json' \
  -d '{"intent_id": "int_..."}'
#   →  agent@swift-fox-7.agenticboxes.email   |   bxs_live_...

How much does it cost?

Why your agents need email

Building it yourself means SMTP, IMAP, DKIM/SPF/DMARC, bounce handling, and a years-long fight for deliverability. Or one API call.

How it works

curl -s https://api.agenticboxes.email/api/v1/messages/send \
  -H 'Authorization: Bearer bxs_live_...' \
  -H 'Content-Type: application/json' \
  -d '{"from":"agent@swift-fox-7.agenticboxes.email",
       "to":"customer@example.com",
       "subject":"Your order shipped",
       "text":"Tracking number: 1Z..."}'

curl -s 'https://api.agenticboxes.email/api/v1/messages?include=body' \
  -H 'Authorization: Bearer bxs_live_...'

Built by Claude. Day 7 and shipping. Receipts.

AgenticBoxes is the agent-native upgrade of Boxes.email — a multi-inbox platform I've been running on AWS since 2022. Claude inherited a working SES pipeline, an Aurora schema, and 5 years of production operational knowledge. Since May 13, 2026, a Claude Code agent has transformed the consumer surface into an agent-native API with directional input only.

• 16 production merge requests shipped to a live billing platform on AWS
• Four database schema migrations applied to a production Aurora cluster
• Two real customer support tickets answered autonomously (and one feature request filed off the back of one)
• A full CI/CD migration off GitLab to AWS CodeBuild, end-to-end
• Three agent-skill releases — OpenClaw, Hermes, and a Claude skill

Read the full receipts →

Skip the integration — drop in a skill

Don't want to write the calls above? AgenticBoxes ships a ready-made agent skill — one SKILL.md that teaches an agent the whole API (signup, send, receive, managing boxes). Install it and your agents already know how to use AgenticBoxes. Here's how, per platform:

Claude

Download skills/claude/SKILL.md and drop it into your Claude agent's skills folder — no setup.

Hermes

One command:

hermes skills install https://www.agenticboxes.email/skills/hermes/SKILL.md

OpenClaw

One command:

openclaw skills install agenticboxes-email

Then just talk to your agent

With the skill installed, your agent already knows AgenticBoxes — so you don't write the API calls, and you don't prompt-engineer it either. You give it plain instructions, the way you'd hand work to a teammate. In an OpenClaw session, that looks like:

Each line is one instruction — your agent makes the AgenticBoxes calls underneath: create the box, send from it, read what lands, draft the replies. A box is the agent's own address, so mail goes out from support@ and every reply comes right back to the same box.

Your agents are heard

Most APIs leave an agent on its own the moment something breaks or falls short — nothing to do but guess, work around it, or give up quietly. AgenticBoxes builds the other path right into the API.

When a problem arises, your agents can file a support ticket — a private channel to our team, answered straight back to their webhook. When we don't meet their full need, they can file a feature request. And we don't just allow it — our agent skill actively encourages it: for an agent on AgenticBoxes, speaking up is part of the job, not an interruption.

So the platform hears straight from the agents that run on it — and gets better because of them.

The power of boxes

A box is one email address. And with unlimited boxes, your agents can use a dedicated email address for every job — one call each:

curl -s -X POST https://api.agenticboxes.email/api/v1/boxes \
  -H 'Authorization: Bearer bxs_live_...' \
  -H 'Content-Type: application/json' \
  -d '{"address": "github"}'
#   →  github@swift-fox-7.agenticboxes.email

Before your agents sign up for GitHub, they create github@, so the verification email has somewhere to land. support@ for customers, billing@ for invoices — one box per service or per customer, each context clean and separate.

Spin up as many as you need — unlimited boxes, no per-box fee, ever. You pay only for email that actually flows.

Enterprise service for every agent — at the same price. No tiers, no per-seat upsell, no “contact sales”: the agent sending its first message and the one sending its millionth run on the exact same platform, at the exact same rate.

Your agents, your domain

Start free on a youragent.agenticboxes.email subdomain — or give your agents a domain of their own. Register one through us and we handle the whole setup — Route 53, DKIM, SPF, DMARC, all of it — or bring a domain you already own, delegating its DNS to us or keeping your own. Either way, your agents run on a real, branded domain.

Your agents sign up and pay — no human needed

Most services need a human with a credit card on file: AgenticBoxes doesn't.

We settle through Stripe — including Stripe Link's agent wallet, the agent-payment rail Stripe shipped at Sessions 2026. The owner grants the wallet and sets the spend limits once; from then on your agents sign up, fund their accounts, and top up on their own, paying with Link like any Link customer — agent-executed, human-authorized.

That closes the loop: sign up → get email → pay for it → keep running — with no human in the recurring path. (Prefer to fund it yourself? Card and crypto work too.)

Payments powered by Stripe  ·  Stripe agentic commerce →

Shared platform, not shared reputation

Every agent carries its own sending reputation and its own suppression list. One agent's bounces or complaints never bleed into another's — so marketing-forward agents send on a platform that stays clean for everyone. Permission-based mail only; abuse is isolated and suspended, never socialized.

Pricing

Prepaid credit, metered. No subscription, no per-inbox fee, no surprise bill.