CosmoBlk/email-marketing-bible

Email marketing skill for AI agents, by George Hartley (founded SmartrMail). Plans, writes, audits and sends email with hard send-safety gates. Works in Claude, ChatGPT, Codex and Cursor. Free, MIT, vendor-neutral.

Email Marketing Bible — Explained

What it is: A free, open-source "skill" file that teaches AI assistants (Claude, ChatGPT, Codex, Cursor, Gemini, etc.) how to do email marketing properly. Think of it as a specialist consultant's playbook that you hand to an AI so it stops making rookie mistakes.

Who made it: George Hartley, co-founder of Nitrosend (an email platform). He previously ran SmartrMail, an email service with ~12,000 customers that sent 6 billion emails before being acquired in 2022. The content is built from 900+ research sources plus that real-world experience.

The core problem it solves: In 2026, AI agents are the ones building campaigns, writing copy, and staging sends — the human just directs. That only works if the AI follows real benchmarks and safety rules. This skill is that discipline layer: it stops an AI from mailing the wrong thing to your whole list, writing generic "AI-slop" copy, or wrecking your deliverability.

What it actually does

  • Audits your email setup — reviews your flows, segments, deliverability, and compliance, then tells you what's missing
  • Builds automated flows — welcome emails, abandoned cart, post-purchase, win-back — from a plain-English prompt, with review before anything goes live
  • Writes copy in your voice — uses proven frameworks (PAS, AIDA, BAB) and strips out the telltale AI phrasing
  • Directs email design — instead of accepting the AI's default template, it generates multiple design directions, scores renders with a "critic loop," and outputs inbox-safe HTML (MJML or React Email)
  • Drives your email platform — connects to services like Klaviyo through MCP/APIs, but with hard safety gates: nothing sends without your explicit approval
  • Fixes spam problems — step-by-step triage covering authentication (SPF, DKIM, DMARC), DNS changes, Gmail's Postmaster tools, and content issues
  • Pulls industry benchmarks — click rates, revenue per recipient, by industry and email type, each with its source and date
  • Compares platforms honestly — by list size, budget, and how well an AI can drive each one (note: the author co-founded one of the compared platforms — disclosed)
  • Checks compliance — GDPR, CAN-SPAM, CASL, UK PECR, Australia's Spam Act, EU AI Act, and more, as a gate before any send
  • Handles cold email, WhatsApp, SMS, and RCS — the real cost models, consent rules, and where each channel makes sense

How you use it

You talk to your AI like you'd talk to a consultant. Examples from the README:

"Audit my Klaviyo account for a DTC skincare brand doing $2M/year. I have a welcome series, abandoned cart, and one weekly newsletter. What am I missing, and build me whatever flow would earn the most first."

"My emails are landing in Gmail promotions and opens dropped from 22% to 14% over three months. What is going on and how do I fix it?"

"Here is a draft welcome email. Make it sound like a person, not an AI."

What's inside

  • 19 chapters covering fundamentals, list building, segmentation, money-making emails, copywriting, design, deliverability, testing, analytics, compliance, industry playbooks, platform choice, cold email, WhatsApp, SMS/RCS, AI & agentic marketing, case studies, expert directory, and 47 curated best email designs
  • Playbooks for 19 industries — ecommerce, SaaS, nonprofits, healthcare, real estate, restaurants, fitness, and more (5 in depth, the rest summarised)
  • Expert contributions — insights from well-known practitioners like Ann Handley, Joanna Wiebe, Nathan Barry, and others
  • A core rules file plus 16 reference files — the AI only loads a reference file when the task needs it, so the rules survive context limits

How to install it

  • Claude Code: add it as a plugin with two commands
  • Other agents (Codex, Cursor, etc.): one command: npx skills add CosmoBlk/email-marketing-bible
  • ChatGPT, Gemini, or any single-file tool: paste the raw core file URL into the chat

Key details

  • Version: 2.8.2, dated 13 October 2026
  • License: MIT (free to use, modify, and share)
  • Full guide online: emailmarketingskill.com, with all chapters, appendices, and a free PDF
  • Contributing: errors, better data, or missing tactics welcome via issues and PRs

A notable design choice

The skill is built around supervised autonomy: the AI can do almost everything on its own, but there's a hard rule that nothing blasts to your list without your say-so. Seed tests to your own addresses (up to five) don't need extra approval; any other single send needs an explicit yes; anything to more than one person waits for "send it."

Related

  • Project
  • Project
  • Project
  • Project
  • Project