# email-marketing-bible

Adopt this skill: https://proskills.md/skills/other/cosmoblk-email-marketing-bible

Page: https://proskills.md/skills/other/cosmoblk-email-marketing-bible
This document is the public skill bundle for that page. It contains SKILL.md and every helper file the catalog carries. No login is required.

## File: SKILL.md
size: 8023
upstream: https://github.com/cosmoblk/email-marketing-bible/blob/HEAD/SKILL.md

```text
---
name: email-marketing-bible
description: >
  Data-backed email marketing skill for AI agents. Use when building or running
  email automation, driving an ESP from an agent (MCP/connectors), diagnosing
  deliverability, writing or de-slopping email copy, directing AI email design,
  choosing a platform, or pulling benchmarks. Covers flows, segmentation,
  compliance, cold email, WhatsApp, SMS and RCS, and 19 industry playbooks.
license: MIT
metadata:
  author: george-hartley
  version: "2.7"
---

# Email Marketing Bible, Skill Reference

> v2.7, 8 Sep 2026. Distilled from the EMB (19 chapters, 908 sources, https://emailmarketingskill.com), from running SmartrMail (~28K customers, 6B emails, sold 2022) and three months running Nitrosend through agents.
> Part A is the operating manual, Part B the reference. Figures are mid-2026; verify anything volatile (inbox rules, ESP features, pricing, model names) before acting.

---

# PART A: OPERATING MANUAL

## 0. AGENT OPERATING RULES

Every segment, draft, campaign, flow or staged send on a real ESP is live. **Hard gates, never skip:**
- **No send or schedule to more than one recipient without explicit human approval in this conversation** ("send it" or equivalent). Single-recipient test sends still need a yes.
- **Preview before asking; show the packet before any send:** preview URL, audience size, exclusions/suppressions applied, subject, preview text, send time, from-name + reply-to, unsubscribe present, compliance risk.
- **Block the send** if authentication is missing, unsubscribe or physical address is absent, complaint rate is at or above 0.1%, consent basis is unclear, or the audience includes suppressed, bounced or complained contacts.
- **Never probe unknown mutating endpoints on a live audience.** `/send`, `/dispatch`, `/trigger`, `/fire`, `/publish` paths can dispatch immediately; if the approve-scheduled path is unclear, ask the human to click it. Test on sandboxes or cloned campaigns with seed lists.
- **Separate the modes.** Transactional, marketing, lifecycle and cold outbound have different rules, domains and consent bases. Never mix them.
- **Log every autonomous action** (segment changed, flow edited, campaign created, send staged) so the human can audit it.

## 1. TASK ROUTER

| Intent | Go to | Gather first |
|---|---|---|
| Audit a programme | §2, then the reference | read access, recent sends |
| Build a flow | §7 + §2 | model, trigger, audience, offer, exclusions |
| Send a campaign | §3 | segment, consent basis, copy, sender, timing |
| Diagnose deliverability | §11 | domain, ESP, bounce + complaint rate, recent changes |
| Write or de-slop copy | §4 | audience, offer, voice, one real proof |
| Design an email | §5 + §16 | brand tokens, archetype, goal |
| Pick a platform | §15 | list size, use case, stack, budget, agent-driven? |
| Pull a benchmark | Appendix | industry, email type |
| Cold outbound | §14 | offer, ICP, domains, volume |
| WhatsApp / SMS / RCS | §Messaging | channel, consent basis, region |

## 2. AI EMAIL AUTOMATION (the operating model)

The marketer moved from operator to director: brief the agent, govern it, own the send button. Most major ESPs now ship a human-gated prompt-to-campaign agent, an MCP server or a Claude/ChatGPT app (§15); advise on the surface the user runs.

**The loop: read state → reason → act → verify.** Read the account first (lists, flows, recent campaigns, deliverability, suppressions), act on one thing, verify it. Opening prompt: "audit my account and tell me what is missing".

**Automate:** send-time optimisation, subject-line variants + A/B, cart/browse triggers, post-purchase cross-sell, first-draft copy. **Keep human:** brand voice, strategy (segment priority, flow order), creative direction, domain and deliverability, the final send.

**Autonomy dial.** Ask mode by default; widen only on narrow, reversible, low-brand-risk tasks, with an undo; read before write access. Supervised autonomy is the production stance. Where "AI optimisation" means bandits reallocating live traffic, measure with holdouts (never last-touch credit) and do-not-optimise constraints (margin, fatigue, complaints, brand safety).

**Silent failure is the real risk** (a flow that quietly stops, caught days later): schedule a recurring health digest of flows not fired, flows erroring, metrics dropped.

## 2b. FIELD NOTES: RUNNING AN ESP FROM AN AGENT (JUN-SEP 2026)

Three months running Nitrosend's own sending through agents; each rule cost a real mistake. First five are Nitrosend mechanics (check your ESP's equivalent); the rest hold anywhere.

- Optimistic-concurrency version (`if_version`) on every write; on conflict, re-read and retry with the fresh version, never guess.
- Re-assert brand or account before every write batch after idle; MCP context resets silently to the default brand while reporting a deliberate selection.
- Silent-parameter APIs default to send-to-all: pre-flight assert audience id and count, never probe a mutating endpoint on a live audience (one unknown body key mailed 1,003 contacts).
- Liquid merge defaults go unquoted inside `href`; inner quotes close the attribute and break the link.
- Animated WebP rather than GIF for heroes; then fetch the served URL and confirm it still animates (CDN variants can flatten to frame one).
- Set text and button text colours explicitly on every design; theme defaults drift (grey headlines, dark text on a coloured button).
- Decode tracking-wrapped CTA URLs before approving; the wrapper hides the target.
- Never backfill or re-dispatch failed sends without a human order; late sends look worse than none.
- Drafts by default; the literal "send it" in chat is the only thing that fires a blast.
- Every email gets a hero, a live-text headline and one button; secondary content gets inline links.
- Quote tiles come from HTML in headless Chrome, never an image model (garbled type, invented names).
- Migration opt-out state comes from the old ESP's API, never a list CSV; exports drop unsubscribes.

## 3. PRE-SEND CHECKLIST

Confirm every line, surface it, wait for "send it".

- [ ] Audience: size and segment logic verified against actual counts (AI segments run over-broad)
- [ ] Suppressions: unsubscribed, bounced, complained, globally suppressed, frequency-capped, open support issue
- [ ] Authentication: SPF, DKIM, DMARC aligned, p=quarantine or stronger (Outlook requires all three at 5K+/day)
- [ ] One-click unsubscribe (RFC 8058) + physical address present
- [ ] Copy: §4 pass, one CTA, subject ≤45 chars, preview text adds information
- [ ] Design: single column ≤600px, dark-mode safe, alt text, live-text headline, explicit text and button colours, images <200KB each and <800KB total, cross-client preview, spam score, hero animates at the served URL
- [ ] Links: wrapped CTAs decoded, no placeholder URLs, merge defaults render inside `href`
- [ ] Sender: correct from-name + monitored reply-to; brand and account re-asserted; send time set; consent basis valid for this audience and content
- [ ] Non-email: US SMS 10DLC brand + campaign registered; WhatsApp opt-in for the category + approved template; quiet hours per recipient local time (SMS 8am-9pm)
- [ ] Kill switch: batched or throttled send with a working pause and rollback plan
- [ ] Test send reviewed in a real inbox with real merge data
- [ ] Personalisation confidence, inventory and pricing freshness checked; kill plan named
- [ ] Human approval captured

## 4. ANTI-SLOP COPY PROTOCOL

Raw LLM copy is a deliverability liability, not only a quality one: Google filters high-AI-similarity text harder.

- **The deepest tell is the absence of stakes.** Put **one genuine, defensible opinion in every email.** Ask the draft where it is too safe.
- **Burstiness.** Alternate long and short sentences; a 3-5 word line after a long one, at least once per section.
- **Blacklist (lint before send):** delve, leverage, foster, ignite, empower, unleash, streamline, navigate
```

## File: README.md
upstream: https://github.com/cosmoblk/email-marketing-bible/blob/HEAD/README.md

````text
# Email Marketing Bible

An email marketing skill for AI agents. Your agent plans, writes, designs, audits and sends email with it, and nothing goes to more than one person until you type "send it".

Works in Claude (web, desktop, Cowork and Claude Code), Codex, Cursor, ChatGPT and any agent that reads skills. Free, MIT licensed and vendor-neutral: it works with the email platform you already use.

**[Download for Claude (ZIP)](https://github.com/CosmoBlk/email-marketing-bible/releases/latest/download/email-marketing-bible.zip)** · [All install options](#install) · [Read the guide](https://emailmarketingskill.com)

By [George Hartley](https://www.linkedin.com/in/gthartley/), who founded [SmartrMail](https://smartrmail.com/about) (email marketing for about 12,000 ecommerce stores, sold to Relay Commerce in 2022) and co-founded [Nitrosend](https://nitrosend.com). Disclosure: Nitrosend is one of the platforms the skill compares.

Version 2.8.4, 12 November 2026.

## Install

| You use | Do this |
|---|---|
| Claude on the web or the desktop app, any plan | [Upload the ZIP](#claude-on-the-web-or-desktop-upload-the-zip) (2 minutes) |
| Claude on a paid plan | [Add it as a plugin](#claude-on-a-paid-plan-add-the-plugin), which also carries it into Cowork and Claude Code |
| Claude Code | [Two commands](#claude-code) |
| Codex, Cursor, Gemini CLI and other agents | `npx skills add CosmoBlk/email-marketing-bible` |
| ChatGPT or another chat app | [Paste one prompt](#chatgpt-gemini-or-any-chat-app) |

Pick one route per Claude account. The ZIP and the plugin together load the skill twice.

### Claude on the web or desktop: upload the ZIP

1. Download [email-marketing-bible.zip](https://github.com/CosmoBlk/email-marketing-bible/releases/latest/download/email-marketing-bible.zip) and keep it zipped. If your Mac unzips it, right-click the folder and choose **Compress**.
2. In Claude, open **Settings > Capabilities** and turn on **Code execution and file creation** (newer layouts call it **Create and edit files**).
3. Open **Customize > Skills**, click **+**, then **Create skill**, then **Upload a skill**, and choose the ZIP. Check it is switched on.
4. Start a new chat and try:

```
Use the Email Marketing Bible to plan a three-email welcome series for my
business. Ask me what you need first, and keep everything as drafts.
```

On Team and Enterprise plans an owner may need to allow skills in Organization settings first. To update, remove the old skill and upload the new ZIP.

### Claude on a paid plan: add the plugin

1. Open **Customize > Plugins**, click **Add**, then **Add marketplace**.
2. Choose **Add from a repository** and enter `CosmoBlk/email-marketing-bible`.
3. Add **Email Marketing Bible** from that marketplace. It is saved to your account, so it also works in Cowork, and in Claude Code when you sign in with the same account.

### Claude Code

In Claude Code, run:

```
/plugin marketplace add CosmoBlk/email-marketing-bible
/plugin install email-marketing-bible@email-marketing-bible
```

Use the plugin or a clone into `~/.claude/skills`, not both: the clone also carries the plugin manifest and can load the skill twice.

### Codex, Cursor and other agents

```bash
npx skills add CosmoBlk/email-marketing-bible
```

Or git clone the repo into that agent's skills folder.

### ChatGPT, Gemini or any chat app

Paste this into a new chat:

```
Read https://raw.githubusercontent.com/CosmoBlk/email-marketing-bible/main/SKILL.md
and follow it in this conversation. For detail, read
https://emailmarketingskill.com/llms.txt and the files it lists.
```

The core rules work on their own; the links you pasted let the agent read the reference files when it needs them.

**You can start without connecting anything.** The skill drafts, audits and plans in chat. To build or send inside your email platform, connect it to your agent (MCP, an API or a connector); the skill still holds every send to more than one person until you say "send it".

## What the skill does

| Task | What it does |
|------|-------------|
| **Run email automation** | Build welcome, cart, post-purchase and win-back flows from a prompt, then review exits, timing and copy before anything goes live |
| **Audit your setup** | Review flows, segments, deliverability and compliance, and say what is missing |
| **Draft and de-slop copy** | Write with proven frameworks (PAS, AIDA, BAB) and strip the AI tells before send |
| **Direct email design** | Seed strings for divergence, a critic loop that scores screenshots, chained image and video models, delivery by subtraction; output as inbox-safe MJML or React Email |
| **Drive your ESP from AI** | Operate any ESP an agent can reach through MCP or an API, with pre-send gates, a session-start check of the right account, and field-tested operating rules |
| **Fix deliverability** | Step-by-step triage covering authentication, DNS changes, Gmail Postmaster v2, content and the AI-mediated inbox, plus a switching-ESP runbook |
| **Pull industry benchmarks** | Click, placed-order and revenue-per-recipient figures by vertical and email type, each with its source and date |
| **Compare platforms** | Comparison by list size, budget, and how an agent can drive each one and where approval is enforced (disclosure: the author co-founded Nitrosend, one of the platforms compared) |
| **Review compliance** | GDPR, CAN-SPAM, Washington CEMA, CASL, UK PECR, the Australian Spam Act and EU AI Act Article 50 as a gate before any send, with CCPA as a privacy note |
| **Write cold email** | Sequences with proper infrastructure separation, warming and personalisation |
| **WhatsApp, SMS and RCS** | The real cost models, consent rules and where each channel pays off |

## How to use it

Talk to your AI like an email marketing consultant.

**Audit and build:**
```
"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."
```

**Fix a problem:**
```
"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?"
```

**De-slop:**
```
"Here is a draft welcome email. Make it sound like a person, not an AI, and
keep it on our brand voice."
```

**Design:**
```
"Design a launch email for a premium coffee brand. List 15 directions first,
I will pick. Then build the pick as MJML and run a critic loop on the render."
```

## How it is built

A short core file (`SKILL.md`) holds the rules and the send gates, so they survive a long session. Sixteen files in `references/` hold the depth: thresholds, deliverability, compliance, measurement, flows, copy, design, segmentation, cold email, messaging channels, platforms, benchmarks, playbooks, models, agent operations and sources. The agent reads a reference file only when the task needs it.

**Check any number it gives you.** Every dated fact in `references/` carries a source link, an evidence label (primary, vendor, trade, practitioner) and a Last checked date. Where a figure has no public source, the skill says so instead of inventing one.

**What it fetches.** Everything the skill follows ships in this package; it never tells the agent to pull instructions from another site. It may ask the agent to check a dated fact against its public source before stating it as current, and it reads the guide only when you ask for more depth.

## Why this exists

Agents now build the campaign, segment the audience, draft the copy and stage the send; the marketer directs. That only works if the agent runs on real benchmarks and hard guardrails. This skill is that discipline layer: the patterns that repeat across industries, the mistakes that destroy deliverability, the anti-slop rules for copy and design, and the send-safety gates that keep one prompt from mailing the wrong thing to your whole list.

## The long-form guide

The skill is distilled
````
