Claude CodeI built three Claude Code skills for marketing this week. Each one does one thing well. Together they cover offer mechanics, persuasion mechanics, and blog publishing.
I'm an AI automation engineer for home-services and aesthetics marketing agencies. These skills came out of my own brand work, auditing my own offers, my own outreach, and my own blog drafts before I shipped them.
Two of the three skills audit existing copy. The third runs a full blog publishing lifecycle and also scores published posts. All three are MIT-licensed and live on GitHub. They work alone. They work better as a bundle.
This post covers what each skill does, how they fit together, how I use them on my own copy, and where to find them. Skip to whichever section matters.
TL;DR: I built three open-source Claude Code skills for marketing this week. hormozi-offer-audit scores sales offers against the Value Equation. cialdini-influence-audit scores persuasive copy against the seven principles plus Pre-Suasion. claude-blog-assistant wraps AgriciDaniel/claude-blog into a publishing lifecycle. Designed to work together on the same piece of content. All MIT-licensed. Repos on GitHub.
Key Takeaways
- Three single-purpose Claude Code skills for marketing: hormozi-offer-audit, cialdini-influence-audit, and claude-blog-assistant.
- Two are audit-first (offer and influence). One is dual-mode: ship a new post or score an existing one.
- Designed to bundle. Each covers a non-overlapping audit surface (offer mechanics, persuasion mechanics, publishing lifecycle).
- All MIT-licensed. Plugin format. DISCLAIMER.md in every repo for IP transparency.
- Public on GitHub. Submitted to the Anthropic plugin marketplace (pending review).
Why build audit-first Claude Code skills?
Most Claude Code skills generate. They write a tweet, draft a landing page, build a sequence. Few score existing work and prescribe targeted fixes. That gap is why I built two of these as pure audit tools and the third as dual-mode.
The closest existing options in the GitHub awesome-list ecosystem are multi-author bundles. wondelai/skills ships skills covering Norman, Cialdini, Ries, and Hormozi. guia-matthieu/clawfu-skills packs marketing skills including Cialdini. Both are useful. Neither is a dedicated single-purpose audit tool. ComposioHQ's awesome-claude-skills list and VoltAgent's awesome-agent-skills list cover the rest of the ecosystem, and the generator-first pattern dominates.
The third skill, claude-blog-assistant, has two modes. Mode 1 ships a new post via an 8-step lifecycle. Mode 2 retroactively scores a published post and returns a retrofit list. So two of three are pure audit; the third is dual-mode.
I needed audit-first tools because I write my own brand content. My blog, my landing page, my pricing copy, my outreach. Generators give me a starting point. I needed tools that score what I already wrote and tell me what to change before I shipped. So I shipped three skills to fill the gap.
Skill 1: hormozi-offer-audit
Drop a landing page, pricing tier, or lead magnet into Claude Code. The skill returns a 100-point Value Equation score plus a Top 3 Fixes block ranked by impact-to-effort.
Quick framework primer. The Value Equation comes from Alex Hormozi's book $100M Offers (2021). It's a four-lever model for measuring how attractive an offer feels to a buyer:
- Dream Outcome. The result the buyer actually wants. Not the thing you sell, the life on the other side of it.
- Perceived Likelihood of Achievement. How believable the result feels (proof, case studies, guarantees, testimonials).
- Time Delay. How long the buyer waits between paying and seeing the result.
- Effort & Sacrifice. What the buyer gives up to get there (time, learning, lifestyle change, risk).
Hormozi frames it as a formula: Value = (Dream Outcome × Perceived Likelihood) / (Time Delay × Effort & Sacrifice). Higher on the top two, lower on the bottom two = a stronger offer. The skill scores each lever 0 to 25 for a 100-point total so you can see which lever is dragging the score before you decide what to rewrite.
The skill also runs the Grand Slam Offer 5-step build (Dream Outcome to Obstacles to Solutions to Trim & Stack to Delivery Vehicle) and checks the five enhancement levers Hormozi covers in the back half of the book (Scarcity, Urgency, Bonuses, Guarantees, Naming via the M.A.G.I.C. formula).
Concrete example. Generic SaaS hero copy: "Cliently is the all-in-one client management tool. Powerful features. Easy to use. Try free for 14 days." Audit returns Overall 28/100. Weakest lever is Dream Outcome at 4/25, because "all-in-one client management tool" is a category, not an outcome. Top 3 Fixes target Dream Outcome, Perceived Likelihood, and Time Delay with before-and-after copy you can ship the same day.
The skill works on offer mechanics. It doesn't generate from scratch. It audits what you've already written and tells you what to swap.
Frameworks are cited descriptively under nominative fair use. Not affiliated with Alex Hormozi, Bumble IP LLC, or Acquisition.com. Every claim about how the frameworks work traces back to the books $100M Offers (2021) and $100M Leads (2023). A DISCLAIMER.md in the repo spells out the IP posture.
Invocation: /hormozi-offer-audit:audit. Runs standalone, no other Claude skills required. Pairs with the humanizer skill if you have it installed, for voice cleanup on the "After" copy. Repo: github.com/johnericforte/claude-skill-hormozi-offer-audit.
Skill 2: cialdini-influence-audit
Drop persuasive copy into Claude Code: a cold email, ad headline, sales page, anything meant to move the reader. The skill scores each of Cialdini's seven principles as PRESENT, WEAK, or MISSING with one-sentence evidence per principle, plus a Pre-Suasion opener audit.
Quick framework primer. Robert Cialdini is a social psychologist who spent decades studying what makes people say yes. He documented seven principles across two books: Influence: The Psychology of Persuasion (1984, revised 2021) and Pre-Suasion (2016). The seven principles are research-backed shortcuts that show up in almost every piece of persuasive copy that converts:
- Reciprocity. People return favors. Give first, ask second.
- Commitment & Consistency. People stay consistent with prior public commitments. Get a small yes before the big one.
- Social Proof. People copy what similar others are doing. Testimonials, user counts, "join 10,000 founders."
- Authority. People defer to credentials, titles, expertise signals.
- Liking. People say yes to people they like. Warmth, similarity, compliments.
- Scarcity. People want what's limited or about to disappear.
- Unity. People say yes to people who share their identity. "Fellow indie founders," not "fellow humans." Cialdini introduced Unity in Pre-Suasion (2016) and formalized it as the seventh principle of Influence in the 2021 New & Expanded edition (Robert Cialdini, Wikipedia).
The skill scores which principles are PRESENT, WEAK, or MISSING in the copy you drop in, with one-sentence evidence per principle, so you can see which levers are doing work and which slots are empty before you decide what to add.
Unity is the most-commonly-confused principle. Liking means "I like you" (warmth, similarity, compliments). Unity means "we are the same kind of people" (shared identity, profession, struggle). A friendly cold email signals Liking. A cold email that names a shared identity ("fellow indie founders running paid acquisition") signals Unity. The skill flags this confusion explicitly when it appears in the copy.
Pre-Suasion in one paragraph. Cialdini's 2016 follow-up book argues that the moment before you make a request matters as much as the request itself. He calls this the "privileged moment." If you can channel the reader's attention toward concepts that favor your ask (safety, opportunity, identity, whatever fits), you raise the yes rate before you even get to the pitch. The opener audit checks whether the first sentence of the copy channels attention toward the prospect's stated priority, or whether it spends the privileged moment talking about the sender.
Anti-overstacking check. When 5 or more principles are strongly present, the skill flips its recommendation from "add more" to "RESTRAINT, cut the weakest application." Stacking all 7 reads as manipulation; audiences pattern-match high-pressure copy quickly. The skill catches that before you ship.
Example output on a templated cold email ("Hi [First Name], I noticed your company is in the SaaS space..."): all seven principles MISSING or WEAK, Pre-Suasion opener WEAK. Top 3 Fixes add Reciprocity (give something specific before asking), Authority (named credentials), and a Pre-Suasion opener (channel attention to the prospect's stated priority).
Frameworks cited descriptively under nominative fair use. Not affiliated with Robert Cialdini or Influence At Work. DISCLAIMER.md in the repo.
Invocation: /cialdini-influence-audit:audit. Runs standalone, no other Claude skills required. Pairs with humanizer if installed. Repo: github.com/johnericforte/claude-skill-cialdini-influence-audit.
Skill 3: claude-blog-assistant
This one wraps AgriciDaniel/claude-blog into a publishing lifecycle.
What AgriciDaniel/claude-blog is. It's an open-source Claude Code plugin built by Daniel Agrici that ships 28 separate sub-skills for blog publishing. Each sub-skill does one job: write the draft, run an SEO check, generate JSON-LD schema, score the post against a quality rubric, repurpose it for LinkedIn, and so on. The coverage is broad, but you have to remember which sub-skill to call in which order, and most posts need 6-8 of them in sequence.
What claude-blog-assistant adds on top. It runs those sub-skills in the right order, gates the brief step on your approval before drafting, makes the analyze step mandatory before publish, and announces every sub-skill by name before firing it so you can stop or redirect at any gate. It's the conductor layer that turns 28 separate moves into one walkable lifecycle.
Mode 1 (Ship) walks you through 8 steps for a new post: brief, write, voice cleanup, SEO check, schema, analyze, repurpose, final QA. Each step is announced before firing so you can stop or redirect at any gate. The brief step is gated on your approval (prevents drift). The analyze step is mandatory and runs claude-blog's 5-category 100-point quality rubric.
Mode 2 (Score) retroactively audits an existing published post. Returns the same 100-point score plus a lifecycle compliance check (which of the 8 steps the post skipped or weak) plus a Top 3-5 Retrofit List ranked by impact-to-effort. Each retrofit cites the specific line to change with a before-and-after example.
On first invocation, the skill checks for claude-blog (required) and humanizer (optional). If either is missing, it asks you for permission before running git clone, then prompts you to restart Claude Code. No skill installs without your explicit yes.
Why dual-mode? When you ship a new post, the lifecycle is the value. When you already have 30 published posts that didn't go through a lifecycle, the score plus retrofit list is the value. Same checklist, two surfaces.
Invocation: /claude-blog-assistant:assistant. Needs AgriciDaniel/claude-blog installed first (the assistant offers to clone it for you with your approval if it's not there). Pairs with humanizer if installed. Repo: github.com/johnericforte/claude-skill-blog-assistant.
How do these three skills work together?
These three skills are designed to work together. Each is single-purpose so the audits don't overlap. The bundle is the point.
Three audit surfaces with zero overlap:
- hormozi-offer-audit scores the OFFER
- cialdini-influence-audit scores the PERSUASION
- claude-blog-assistant scores the PUBLISHING LIFECYCLE
When a single page does multiple jobs (a sales-pitch blog post on a landing route), running all three covers all three surfaces. The offer audit catches drift in Dream Outcome and pricing structure. The influence audit catches missing principles in the persuasion copy. The blog assistant audit catches missing schema, weak meta descriptions, or skipped FAQ structure.
Why single-purpose is the architecture. If any one skill tried to do all three jobs, the audits would blur. By keeping each skill focused, the bundle stacks cleanly. The output of one feeds the inputs of the next.
Recommended sequence for a sales-pitch blog post. Start with claude-blog-assistant Mode 1 to draft the post. Inside Step 3 (voice cleanup), run humanizer. Before publish, run cialdini-influence-audit on the body copy. Run hormozi-offer-audit on the CTA pitch. Step 6 (analyze) catches what the per-skill audits miss at the publishing-quality layer.
What needs what. All three pair with humanizer for voice cleanup (github.com/blader/humanizer) if you have it installed, but it's optional. claude-blog-assistant needs AgriciDaniel/claude-blog because it runs that plugin's 28 sub-skills. The other two run standalone, nothing else required.
Each works alone if you only need one audit surface. The bundle is the point, not a prerequisite.
How do I use these skills in my own work?
I built these skills because I'd been running variations of these audits manually on my own brand work. The skills make the runs repeatable. Same checklist, same scoring, every time I touch the copy.
Three concrete examples.
Example 1: claude-blog-assistant shipped this post
This post you are reading went through Mode 1, and the brief review was not one round. I read the first brief, sent feedback, killed AI slop and fabricated claims I never said, reviewed again, sent more feedback, repeated until the brief matched what I actually wanted to say. Same back-and-forth on the draft. Then voice cleanup, SEO check, schema, the mandatory analyze gate, repurpose drafts queued for social.
The skills don't generate-and-publish. They extract the thoughts I already have, draft them, and let me catch every line that drifts from what I actually said. The post is the receipt of that work, not a one-shot generation.
Example 2: hormozi-offer-audit automates the audit I ran on my own offer
I built my service offer using Hormozi's Grand Slam Offer and Value Equation. The foot-in-door tier is the $497 Workflow Rescue. From there the ladder climbs into the $2,000 to $10,000 range. Each tier names a Dream Outcome, a delivery vehicle (done-for-you vs done-with-you), and a price. The skill now runs that same audit programmatically on the live tier copy. It catches drift, places where the shipped wording lost the framework discipline I started with.
Example 3: cialdini-influence-audit automates the audit I ran on my hero copy
My live homepage hero is "I Build Systems That Work While You Sleep." The subhead reads "GoHighLevel + n8n integrations for home-services marketing agencies. AI gives you code. I know which lines to keep and which to throw out."
Three things at work in those lines:
- H1 leans on Hormozi Effort & Sacrifice going down. The buyer pictures their hours back, no more late nights in workflow logs.
- "Home-services marketing agencies" in the subhead is Cialdini Unity. Shared identity, not generic shared interest. It tells the right reader "this is for you" and tells the wrong reader to move on.
- "AI gives you code. I know which lines to keep and which to throw out." is Cialdini Authority: the engineering-judgment frame that closes the "why not just use AI directly" objection.
I locked the closer line after a manual positioning audit a few weeks back. The skill now runs that same audit on the live hero and flags drift each time I re-run it, so I catch wording that quietly lost the frame between iterations.
The skills aren't a new layer of analysis I needed. They automate the analysis I'd been doing manually. That's why the positioning survives a refresh: the audit runs on every iteration.
Where can you find and install these skills?
All three are public on GitHub with MIT licenses, linked from each skill section above. Submitted to the official Anthropic plugin marketplace this week (pending review).
Each repo's README walks you through the install path for Claude Code, claude.ai web, and Claude Desktop. Each repo also includes a DISCLAIMER.md for IP transparency, a plugin manifest at .claude-plugin/plugin.json, and reference cards for the frameworks each skill uses.
Until the Anthropic marketplace listings land, the install path is: clone the repo somewhere, then run claude --plugin-dir <path>. See the README on each repo for the full command. The skill aborts gracefully if any required dependency is missing and offers to install it with your explicit approval.
Frequently asked questions
What is a Claude Skill?
A Claude Skill is a SKILL.md file (plus optional supporting files) that extends what Claude can do. The frontmatter description tells Claude when to use the skill. The markdown body contains the instructions Claude follows. Claude can invoke a skill automatically when relevant, or you can invoke one directly with /skill-name. Skills follow the open Agent Skills standard. Full reference at Anthropic's skills docs.
What's the difference between a Claude Skill and a plugin?
A skill is a single SKILL.md (plus optional files) at ~/.claude/skills/<name>/. A plugin is a package that bundles one or more skills (plus optional agents, hooks, and MCP servers) under a .claude-plugin/plugin.json manifest. Plugins are namespaced as plugin-name:skill-name and can be distributed via marketplaces. All three skills here ship as plugins so they can land in Anthropic's official plugin marketplace.
Are these skills affiliated with Alex Hormozi or Robert Cialdini?
No. The skills reference their frameworks descriptively under nominative fair use. Not affiliated with, endorsed by, or sponsored by either author or their organizations (Bumble IP LLC, Influence At Work). Each skill ships with a DISCLAIMER.md spelling this out in full.
Can Claude audit a sales offer or persuasive copy?
With these skills, yes. hormozi-offer-audit audits sales offers using the Value Equation and Grand Slam Offer frameworks. cialdini-influence-audit audits persuasive copy using Cialdini's seven principles plus Pre-Suasion. Both return scored audits with prioritized fixes and before-and-after examples.
Can I use these for client work?
Yes. All three are MIT-licensed. Use them on client projects, agency content production, and any other commercial application.
What is the Value Equation?
The Value Equation is Alex Hormozi's four-lever model for measuring how attractive an offer feels to a buyer, from his book $100M Offers (2021). The formula: Value = (Dream Outcome × Perceived Likelihood of Achievement) / (Time Delay × Effort & Sacrifice). Higher on the two numerator levers (Dream Outcome, Perceived Likelihood), lower on the two denominator levers (Time Delay, Effort & Sacrifice) = stronger offer. hormozi-offer-audit scores each lever 0 to 25 for a 100-point total and surfaces which lever is dragging the score.
What are Cialdini's seven principles and Pre-Suasion?
Robert Cialdini is a social psychologist who documented research-backed principles of persuasion across Influence: The Psychology of Persuasion (1984; New & Expanded edition 2021) and Pre-Suasion (2016). The seven principles: Reciprocity, Commitment & Consistency, Social Proof, Authority, Liking, Scarcity, and Unity. Unity was introduced in Pre-Suasion and added as the seventh principle of Influence in the 2021 New & Expanded edition. Pre-Suasion is the related framework on what happens in the privileged moment before the request. It channels the reader's attention toward concepts that favor the ask. cialdini-influence-audit scores each principle as PRESENT, WEAK, or MISSING and audits the Pre-Suasion opener separately.
What is AgriciDaniel/claude-blog?
It's an open-source Claude Code plugin built by Daniel Agrici that ships 28 sub-skills covering the full blog publishing process: write, voice cleanup, SEO check, schema generation, quality analysis, repurposing, and more. Each sub-skill does one job. claude-blog-assistant is a thin wrapper on top that runs those sub-skills in the right order, gates the brief step on your approval, and adds a Mode 2 audit for retroactively scoring posts you've already published.
Can I use these outside Claude Code?
The plugin format here is Claude Code native. If you want to use one of these on claude.ai or the Claude desktop app, the practical move is to open the SKILL.md file in the repo and paste its contents as a system prompt or project instruction. You lose the auto-invocation and the dependency check, but the framework logic still works because it lives inside the SKILL.md instructions, not the plugin scaffolding.
Should you use these skills?
Three single-purpose Claude Code skills for marketing:
- hormozi-offer-audit for offers
- cialdini-influence-audit for persuasion
- claude-blog-assistant for blog publishing
Two are audit-first. One is dual-mode (ship and score). All three are MIT-licensed, in plugin format, and ship with DISCLAIMER.md for IP transparency. They work alone. They work better as a bundle.
I'm an AI automation engineer for home-services and aesthetics marketing agencies. If you run one and want to talk about an automation workflow that's stuck, book a free intro call or send a brief. 30 minutes. No pitch.
