Skip to content
Christian Monge

Claude Skills

Claude Marketing Skills: What to Install and Build

Claude marketing skills worth using in 2026: the best open-source library, how to install it, which skills to start with, and how to build gates of your own.

cmChristian Monge4 min read
A miniature marketing studio of frosted glass objects, a megaphone, a bar chart and a camera, linked by thin electric blue threads

Claude marketing skills turn marketing frameworks into instructions Claude applies automatically: an SEO audit checklist, a copywriting process, a pricing framework. The best free starting point is Corey Haines' open-source marketingskills library. The biggest gains, though, come from the skills you write around your own rules.

I'm a marketer who ended up building software, so I use these skills on real work: my own site, my family's e-commerce store and client content. Here's what I'd install first, and what I'd build myself.

What makes a marketing skill worth installing?

A skill is a folder with a SKILL.md file that Claude loads when a request matches its description. If you're new to the format, read my guide to Claude skills first.

For marketing, I judge a skill on three things:

  1. Does it read context before acting? A copywriting skill that doesn't know your audience writes generic copy.
  2. Is it specific? "Write a good landing page" is a prompt. A skill should carry a real process, like an ordered CRO checklist.
  3. Can it say no? Most marketing skills generate. The valuable ones also review, and flag or block work that misses a standard.

The best library: marketingskills by Corey Haines

The marketingskills repository is the most complete public collection I've found. It's MIT licensed, maintained actively, and works in Claude Code, OpenAI Codex, Cursor, Windsurf and any tool that supports the Agent Skills format. As of September 2026 it had close to 50,000 GitHub stars.

Its smartest design choice is a shared foundation. The product-marketing skill creates a context document about your product, audience and positioning, and every other skill reads it before doing anything. That's exactly what generic AI marketing output is missing.

The skills are grouped like a marketing team:

AreaExample skills
SEO and contentseo-audit, ai-seo, schema, site-architecture, programmatic-seo, content-strategy
Conversioncro, signup, onboarding, popups, paywalls
Copy and contentcopywriting, copy-editing, cold-email, emails, social, video
Paid and measurementads, ad-creative, ab-testing, analytics, attribution
Growth and retentionreferrals, free-tools, lead-magnets, churn-prevention
Sales and go-to-marketlaunch, pricing, competitors, sales-enablement, prospecting
Strategymarketing-ideas, marketing-psychology, customer-research

One transparency note: the library lists paying tool partners in a clearly labelled section. The maintainer states that sponsorship never changes what the core skills recommend, but read integration guides with that in mind.

Claude Code: Build Your Full AI Marketing Team (Agents + Claude Skills) · Grace Leung

How to install Claude marketing skills

In your project folder, run:

npx skills add coreyhaines31/marketingskills -a claude-code

The -a claude-code flag matters. The installer detects your agents, and if it runs non-interactively it may only write to a universal .agents/skills/ folder, which Claude Code doesn't read. With the flag, the files land in .claude/skills/.

To install only a few skills:

npx skills add coreyhaines31/marketingskills --skill product-marketing seo-audit copywriting -a claude-code

Then open Claude Code and run the foundation first: ask Claude to create your product marketing context. Answer its questions properly. Every later skill is only as good as that file.

The five marketing skills I'd start with

If you install everything at once, you won't learn what any single skill does. Start with these, in this order:

  1. product-marketing. The context file everything else depends on.
  2. seo-audit. Point it at your key pages and get a prioritized list of technical and on-page issues.
  3. ai-seo. Search increasingly happens inside AI answers. This skill focuses on getting cited by LLMs, not only ranked by Google.
  4. copywriting. Use it on one page you already know converts poorly, then compare the before and after with real data.
  5. customer-research. Feed it interviews, reviews or support tickets and let it pull out the language customers actually use.

Run each one on a real asset, check the output against what you know, and only then add the next.

Build the skills no library can give you

Public skills know marketing. They don't know your business. The most valuable marketing skills I use are ones I wrote, because they hold rules that only exist in my work.

The clearest example: on one live site, the rule "every post must link to the product" was written in three separate project documents. When I finally measured it, 45 of 50 published posts broke it. Nobody ignored the rule on purpose. Nothing checked it.

So my own content skill doesn't write. It audits. It runs a script with 38 on-page rules, covering title length, a declared keyword, internal links, sourced statistics and FAQ structure, and returns an error code when a draft fails. Claude can't talk its way past an exit code.

Here's the kind of custom marketing skill worth building:

  • Brand voice gate: banned phrases, required proof points, tone rules, checked before anything publishes.
  • Keyword collision check: stops a blog post from targeting a keyword your sales page already owns.
  • Claim verifier: confirms every statistic in a draft actually appears on the cited source.
  • Campaign naming check: rejects ad campaigns that don't follow your reporting convention.

Keep the rules that must be exact in scripts, and keep SKILL.md short. Anthropic's skill format is built for this: scripts run and only their output enters Claude's context.

Skills alone aren't a marketing system

Skills handle procedures. A real AI marketing setup also needs memory: past campaigns, what worked, customer language, decisions and why they were made. That's where a knowledge base comes in, and it's what I build with Obsidian and Claude Code. If you want to see how that fits together, read how to build an Obsidian second brain with Claude Code, or look at the AI Brains I build for marketing teams.

And if you're unsure whether a job belongs in a skill, a subagent or an MCP connection, my breakdown of skills vs agents vs MCP makes the choice concrete.

Frequently asked questions

What are the best Claude skills for marketing?

For breadth, the open-source marketingskills library by Corey Haines covers SEO audits, AI search, copywriting, CRO, email, ads, analytics and pricing. For depth, pair a few of those with custom skills that hold your own positioning, thresholds and review rules.

How do I install marketing skills in Claude Code?

Run npx skills add coreyhaines31/marketingskills -a claude-code to install the whole library into .claude/skills/, or add --skill with specific names to install only the ones you need.

Do Claude marketing skills replace a marketer?

No. They package frameworks and checklists so Claude applies them consistently, but they don't know your customers, your numbers or what worked last quarter. That context still has to come from you.

Are there Claude skills for SEO?

Yes. The marketingskills library includes seo-audit, ai-seo, schema, site-architecture and programmatic-seo, and there are dedicated SEO skill collections for Claude Code as well.

Sources

  1. 01coreyhaines31/marketingskills (GitHub)
  2. 02Agent Skills overview (Claude Platform docs)
  3. 03anthropics/skills repository (GitHub)
  4. 04What are skills? (Claude Help Center)
cm

Christian Monge

Builds AI second brains with Obsidian and Claude Code, co-founded the competitive-intelligence SaaS Pengu Insights, and trains teams in English and Spanish from Madrid.

More about me

AI Brains

Want this running on your own material?

I build Obsidian + Claude Code brains around your documents, calls and research, with an evidence ledger so every number has a source.

See the AI Brains

Keep reading