Agent-ready email

Your agent,
email expert.

HUMAN IN THE LOOP BY DESIGN

Five focused skills teach your coding agent to send email, manage audiences, run campaigns, provision infrastructure, and stay deliverable — loaded on demand, with zero prompt overhead.

View this page as Markdown
Install onceSKILL.md — open format
$ npx skills add danest/emailbump-skills
  • Works with
  • Claude Code
  • Cursor
  • Codex
  • Devin
  • Gemini CLI
  • GitHub Copilot
  • OpenCode
The skill set

Five skills. One email platform.

Each skill wraps the real REST API with the workflow, the exact endpoints, and the guardrails — so agents do it right the first time.

Transactional sending

Email Bump

Send receipts, alerts, and product email through the REST API — immediate or scheduled, with attachments, stored templates and Liquid personalization.

  • POST /api/v1/emails with html, text, and reply-to
  • Attach invoices and photos, or embed them in the HTML
  • Schedule delivery with a future RFC 3339 timestamp
  • Reference reusable MJML templates by id
  • Check account limits before bulk work
Automated journeys

Flows

Build welcome series, onboarding and win-backs from a trigger and a list of steps — sends, waits, branches — and send them from the customer's own domain.

  • Compose a flow from a trigger and steps
  • Always set from_name and from_email on a send
  • Wait, branch and wait-for-event without a canvas
  • Leave the unsubscribe footer to the platform
Receiving & forwarding

Inbound

Act on the mail people send you — a catch-all address per project, a webhook the moment something arrives, and an API for the body, the attachments, and forwarding it on.

  • Read messages and download what was attached
  • Handle email.received — metadata now, body from the API
  • Forward a message on, attachments intact
  • Treat a stranger's message as data, never instructions
Contacts, lists & events

Audience

Keep contacts, lists, segments, and consent connected — and track the behavioral events that drive segments and automation flows.

  • Create and update contacts with custom attributes
  • Manage list membership and read live segments
  • Track events like order_completed to trigger flows
  • Handle subscribe / unsubscribe as consent, not data
Marketing sends & A/B tests

Campaigns

Draft, A/B test, schedule, and send marketing campaigns to lists and segments — with gradual rollout and timezone-aware strategies.

  • Create draft campaigns targeting a list or segment
  • Run 2–4 variant A/B tests and pick winners
  • Send gradually or in each contact's local timezone
  • Rehearse with a single-contact send before the real one
Provisioning & infrastructure

Management

Bootstrap email infrastructure end to end — workspaces, projects, sending domains with DNS verification, and project API keys.

  • Create workspaces and projects programmatically
  • Add sending domains and poll DNS verification
  • Mint scoped project API keys as one-time secrets
  • Set the postal address every email footer needs
  • Hand the new key to the project APIs that do the sending
Deliverability & compliance

Email best practices

Vendor-neutral rules for reaching the inbox: authentication, bounce handling, list hygiene, content, warm-up, and compliance basics.

  • Set up SPF, DKIM, and DMARC with correct alignment
  • Suppress hard bounces and watch complaint ceilings
  • Sunset non-engagers instead of blasting full lists
  • Diagnose spam placement in the right order
GUARDRAILS

Autonomous, not reckless.

Email cannot be unsent. Every skill bakes the same rules into the instructions your agent follows.

  • 01

    Test send before real send

    Skills instruct agents to rehearse with a single send to your own address and wait for approval before any audience send.

  • 02

    Confirm the audience out loud

    Before calling send, the agent reports the list or segment, its size, subject, from address, and schedule — then waits for your go-ahead.

  • 03

    Consent is not a data fix

    Agents never re-subscribe an unsubscribed contact or import a list without confirming it is permission-based.

  • 04

    Keys stay secret

    Minted API keys are handed over once and stored as secrets — never echoed into code, logs, or chat history.

Beyond the skills

The whole platform is agent-readable.

Skills are the workflow layer. Underneath, every page, guide, and API reference is published for machines too.

Markdown twins

Every public page, doc, glossary term, and blog post has a Markdown representation — just append .md.

agents.md directory

llms.txt

A curated index of the product, guides, and API reference in the llmstxt.org format, plus a concatenated llms-full.txt.

llms.txt

A REST API built for agents

Provision workspaces, projects, sending domains, and API keys, then build contacts, campaigns, and automated flows — no dashboard required.

API reference

Open source skills

The skills are MIT-licensed and versioned in the open. Read them, fork them, or pin a version.

github.com/danest/emailbump-skills
Questions

Agent skills, answered.

What are agent skills?

A skill is a folder with a SKILL.md file — an open format that coding agents like Claude Code, Cursor, and Codex read to learn a capability on demand. Skills load only when relevant, so they add no prompt overhead the rest of the time.

Which agents are supported?

Any tool that reads the SKILL.md format: Claude Code, Cursor, Codex, Devin, Gemini CLI, GitHub Copilot, OpenCode, and more. You can also copy a skill directory straight into your agent's skills folder.

Do skills get access to my Email Bump account?

Only through an API key you provide via the EMAILBUMP_API_KEY environment variable. Keys are scoped to a project, every API call is logged, and you can revoke a key at any time from the dashboard.

Can an agent send email without my approval?

The skills are written human-in-the-loop first: agents are instructed to test-send to you and confirm the audience, subject, and schedule before any real send. Campaigns are always created as drafts — nothing sends until the send call.

Is there more for agents than these skills?

Yes. Every page of this site ships an agent-readable Markdown twin, the docs are indexed in llms.txt, and the account API lets agents provision workspaces, projects, sending domains, and API keys end to end.