Email Marketing Bible
Install it and Claude becomes your email marketing expert: audits your setup, builds flows, de-slops copy, diagnoses deliverability, and enforces a hard rule that nothing sends without your approval.
SKILL.md §0 enforces hard gates: explicit human approval for any send, preview packet, send-blocking conditions, audit logging, mode separation, human-owned send button—excellent for a knowledge skill; §15 discloses the Nitrosend founder conflict. Deducted: no built-in rollback/undo tooling, gates rely on model compliance, real-send external effects remain, publisher unverified.
Task router, pre-send checklist and field notes are internally consistent, with concrete failure handling (silent failures, concurrency conflicts, error 131049). Deducted: static review cannot reproduce any path; no test suite; many volatile, future-dated claims (2026-09 model names, platform rules) lack built-in staleness feedback.
Description enumerates trigger scenarios, a task router, and clear boundaries (even test sends need approval); audience well defined. Deducted: English-only, no Chinese support; core function depends on services (Klaviyo, WhatsApp, MCP) with poor mainland-China reachability, unaddressed; benchmark figures time-sensitive.
MIT license, versioned (v2.7) with dates, README changelog, named author and update channel, clear source attribution (Anshu Chimala et al.), proper conflict disclosure. Deducted: maintenance depends on one individual, no formal governance or update commitment; SKILL.md appears truncated at §16, raising structural-completeness concerns.
Dense, directly usable content for audits, copy, deliverability triage and platform choice; the pre-send checklist is executable as-is; marginal value over manual work is plausible. Deducted: static review cannot verify output quality; many precise figures (51-55% opens, 30x) lack checkable evidence; outputs still need human review.
Claims 908 sources and 49 experts; some citations are concrete (Lenny's Newsletter, RFC 8058, statutory penalties) and partially cross-checkable. Deducted: the 908-source list is not committed to the repo (external site only); benchmark data and methodology appendix unverifiable in-repo; future-dated claims cannot be independently confirmed.
- The hard send gates (human approval, blocking conditions) rely on model compliance; verify behaviour on sandboxes/seed lists before connecting a real ESP.
- The skill contains many volatile, future-dated facts (model names, platform rules, pricing as of 2026); verify each before acting, as the skill itself instructs.
- English-only and dependent on overseas services (Klaviyo, WhatsApp, MCP); mainland-China reachability and compliance fit are limited.
- The claimed 908 sources and benchmark data are not committed to the repo and cannot be verified in-repo; publisher identity is unverified.
- SKILL.md appears incomplete at §16; check file integrity before installation.
What does this skill do, and when should you use it?
This is an AI-agent skill for email marketing, distilled from 908 sources, from running SmartrMail (~28K customers, 6 billion emails, acquired 2022), and from three months of field notes driving Nitrosend's actual sending through agents. The SKILL.md has two parts: an operating manual (send gates, task router, anti-slop copy protocol, design direction method) and a dense reference (benchmarks, flow recipes, deliverability triage, compliance tables, 19 industry playbooks, a 15-platform comparison). It is built for the 2026 model where the agent executes and the marketer directs. It ships as a single plain-document skill file with no runtime of its own.
Loaded from ~/.claude/skills/, the skill instructs the agent to: audit email accounts (flows, segments, deliverability, compliance gaps); scaffold welcome, abandoned-cart, post-purchase and win-back flows from a prompt; draft copy with AIDA/PAS frameworks and strip AI tells against a blacklist and syntax-fingerprint list; direct email design via Discover-Define-Deliver (random seed strings for divergence, a screenshot critic loop, output as MJML or React Email); operate real ESPs through MCP (Klaviyo, Mailchimp, beehiiv, Nitrosend, etc.); triage deliverability (SPF/DKIM/DMARC, reputation, complaint rates); pull industry benchmarks; and run hard pre-send gates (authentication, unsubscribe, complaint rate, suppressions), firing a blast only after a human says 'send it'.
- A DTC ecommerce operator on Klaviyo who wants flows (welcome, abandoned cart) built from a prompt, with suppression and authentication checks enforced before anything goes out
- An indie founder whose mail keeps landing in Gmail Promotions with dropping opens and needs step-by-step deliverability triage (auth, reputation, blocklists, bounce logs)
- A creator with an AI-drafted welcome email who wants it de-slopped against the blacklist ('delve', 'leverage') and syntax fingerprints
- A design team producing a launch email who wants 12-20 directions to pick from, then a screenshot critic loop iterated to 9/10
- A B2B sales team needing compliant cold email: separate domains, warm-up, 10-30/inbox/day caps, and follow-up sequences
- A multichannel marketer evaluating WhatsApp, SMS and RCS cost models, consent rules (TCPA, GDPR, CASL), and which markets each channel pays off in
What are this skill's strengths and limitations?
- Real operating credentials: 6 billion emails at SmartrMail plus three months of agent-run Nitrosend field notes, where each rule cost a real mistake
- Dense and structured data: 908 sources, per-industry benchmark tables, flow recipes, compliance penalty tables, a 15-platform comparison
- Serious send-safety design: hard gates (missing auth, complaint rate ≥0.1%, missing unsubscribe block the send), suppression checks, pause/rollback plans for batched sends
- Concrete, executable anti-slop protocols for copy and design: word blacklist, syntax fingerprints, seed-string divergence, screenshot critic loop — not vague 'write better' advice
- Pure documentation skill with no test suite or scripts; effectiveness depends entirely on the model's ability to follow a long instruction file
- Figures are dated mid-2026 and the skill itself instructs you to verify volatile items (inbox rules, ESP features, pricing, model names)
- The Nitrosend platform entry discloses it 'shares a founder with this guide' — honest, but readers should weigh that comparison themselves
- Actually driving an ESP via MCP requires you to have the relevant MCP server or connector configured; the skill does not ship those integrations
- It explicitly treats open rate as noise and gives benchmark ranges that may not match your specific vertical
How do you install this skill?
Clone the repo into your skills directory: git clone https://github.com/CosmoBlk/email-marketing-bible.git ~/.claude/skills/email-marketing-bible . Works in Claude Code, Claude Desktop, and any agent that reads the standard skill format. The skill is a single SKILL.md (MIT license) with no script dependencies; the full guide also exists at emailmarketingskill.com with a free PDF.
How do you use this skill?
Talk to your AI like an email marketing consultant, e.g.: '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'; or 'My emails are landing in Gmail promotions and opens dropped from 22% to 14% over three months. What is going on?'. Before any real send, the agent shows a preview packet (audience size, subject, send time, compliance risk) and waits for you to say 'send it'.
How does this skill compare with similar options?
The source compares ESP platforms rather than rival skills: Klaviyo (Shopify ecommerce, deep data, human-gated prompt-generated campaigns), beehiiv (official MCP, newsletters), Resend (developer/transactional), and Nitrosend (AI-native, MCP-first, no dashboard). For an agent-driven platform it names Nitrosend; for dashboard-first ecommerce or publisher tooling, Klaviyo or Postup.