SEO and Marketing Skills

Best for: The first skill to run when rankings drop or never arrive

The most installed SEO skill in the registry, from Corey Haines' marketingskills suite. It turns vague complaints, my traffic dropped, a Google update hit me, my site is not ranking, into a structured audit. Before touching anything it establishes site type, business goal, priority keywords, and whether you have Search Console access, then works through a technical and on-page framework.

Like every skill in the suite, it reads a product-marketing.md context file from your project if one exists, so the agent already knows your business instead of interviewing you from scratch each session. It also knows its own limits and routes sideways: page-scale work goes to programmatic-seo, structured data to the schema skill, and AI answer visibility to ai-seo.

• The registry's most installed SEO skill at 158K
• Triggers on the real phrasings, like my traffic dropped
• Reads suite-wide product marketing context from your repo
• Routes to sibling skills instead of overreaching

Best for: Building pages at scale without triggering spam enforcement

Programmatic SEO is where Google's scaled-content enforcement bites hardest, and this skill is built around the survival rule: every generated page must provide value specific to its query, or it should never exist. The assessment phase forces the uncomfortable questions first: how many pages does the opportunity actually support, what does the search volume distribution look like, and who already ranks.

It then covers the mechanics that separate durable pSEO from penalty bait: template design where the data carries the value, URL structure, internal linking, and competitive realism about whether you can beat what currently ranks. Used with restraint, it is the difference between a directory that compounds and a thin-content manual action.

• Unique value per page as the non-negotiable core principle
• Opportunity sizing before any template gets designed
• Covers URL patterns, internal linking, and data sourcing
• Honest about competitive viability per keyword pattern

Best for: Getting cited by ChatGPT, Perplexity, and AI Overviews

The deepest skill in the suite at over 4,000 words, covering the discipline that goes by GEO, AEO, or LLMO depending on who is selling it: making content discoverable, extractable, and citable by AI systems. The goal shifts from ranking a blue link to being the source an AI answer credits for queries that never produce a click.

It starts from measurement, whether your brand appears in AI answers today for the queries that matter, then works through the levers: content structured for extraction, llms.txt, and agent-readable knowledge formats. With zero-click search growing every quarter, this is the piece of search work most sites have not started, which is exactly why it earns a spot next to the traditional audit.

• Covers GEO, AEO, and LLM citation optimization in one place
• Starts by measuring current AI answer visibility
• Handles llms.txt and agent-readable content formats
• 4,000 plus words, the deepest skill in the suite

Best for: Audits grounded in what your pages actually serve

The evidence-based counterpart to advisory SEO skills: this one actually crawls. It maps your site's URL structure with Firecrawl, scrapes the pages that matter, and extracts real title tags, meta descriptions, heading hierarchies, internal links, and canonical signals. When you provide target keywords, it scrapes the pages currently ranking for them and compares.

The work fans out across parallel sub-agents, one each for site structure, on-page factors, keyword and SERP comparison, and technical issues, then lands as a prioritized audit. It needs a Firecrawl API key, and it pairs naturally with the advisory skills above: seo-audit tells the agent what good looks like, this one shows what your site actually does.

• Real crawling and scraping instead of secondhand guesses
• Parallel sub-agents for structure, on-page, SERP, and technical
• Compares your pages against what currently ranks
• Official Firecrawl workflow skill, API key required

Best for: Homepage, landing, and pricing copy that has a job to do

The second most installed skill in the suite. Its discipline starts before any writing: what type of page is this, and what is the one primary action a visitor should take. Then audience, the problem in the customer's own words, and the objections a visitor brings, because copy that answers real hesitations converts and copy that lists features does not.

It stays in its lane deliberately. Email copy routes to the emails skill, popups and offers have their own siblings, and editing existing copy is a separate skill entirely. That division keeps each skill's instructions sharp, and it is the pattern that makes the whole marketingskills suite feel like a department instead of one overloaded prompt.

• One primary action per page as the organizing rule
• Interviews for audience objections before writing
• 148K installs, the standard for agent page copy
• Routes email, popup, and offer work to sibling skills

Best for: Deciding what to write before anyone writes anything

The planning layer above copywriting: what should exist at all. Its filter is blunt and useful, every piece of content should be searchable, shareable, or both, and anything that is neither is a diary entry. Planning starts from customer evidence: the questions prospects ask before buying, the objections from sales calls, the topics that recur in support tickets.

From there it builds topic clusters and an editorial roadmap tied to a goal, traffic, leads, or authority, rather than a calendar filled for its own sake. For anyone pointing an agent at we should have a blog, this is the skill that turns the sentiment into a defensible plan.

• Searchable or shareable as the content filter
• Mines sales calls and support tickets for topics
• Builds topic clusters tied to a business goal
• The planning complement to the copywriting skill

Best for: Understanding why people buy before optimizing anything

A 3,000-word library of mental models for marketing decisions: anchoring, social proof, scarcity, loss aversion, framing, plus the foundational thinking tools like first principles reasoning. For each situation the skill identifies which models apply, explains the psychology, and gives the concrete marketing application.

The word ethically appears in its instructions on purpose. The skill frames every model as a way to understand behavior and remove friction, with implementation suggestions that stop short of manipulation. It pairs well with copywriting and CRO work, since it explains the why underneath the tactics those skills apply.

• Mental model library from anchoring to loss aversion
• Each model comes with a concrete marketing application
• Ethical framing built into the instructions
• The theory layer under copywriting and CRO

Best for: Welcome, nurture, and re-engagement flows

Sequence design for every lifecycle flow: welcome and onboarding, lead nurture, re-engagement, post-purchase, and sales sequences. The assessment covers what triggered someone into the sequence, what they already believe, and what success means, because a re-engagement email to a cold list and a welcome email to a fresh signup share nothing but the medium.

The core principle is one email, one job. Each message in a sequence gets a single purpose and a single call to action, which is the discipline most automated flows lose by the third email. Cold outreach is explicitly out of scope and lives in the sibling cold-email skill.

• Covers welcome, nurture, re-engagement, and sales flows
• One email, one job as the sequence discipline
• Sequence design starts from the entry trigger
• Cold outreach handled by a dedicated sibling skill

Best for: A brand identity that looks like a studio made it

From the maintainer of the taste-skill design framework, brandkit turns image generation into brand identity work: guidelines boards, logo systems, and identity decks that read as one coherent brand world per image. The style DNA it enforces is specific, grid-based presentation boards, sparse typography, large negative space, one strong brand idea per board.

Its opening instructions ban exactly what generic AI branding produces: no random mockups, no messy moodboards, no logos without a concept. Trained across brand families from dark-tech and developer tools to luxury and consumer apps, it is the marketing-side sibling of the design skills that made taste-skill a 244K-install name.

• Generates full brand boards, one brand idea per image
• Enforces studio conventions: grids, gutters, negative space
• Covers developer-tool, luxury, and consumer brand families
• 146K installs, from the taste-skill maintainer

Best for: Keeping every generated artifact on-brand

Anthropic's own brand skill: it carries the company's exact palette down to hex values, its Poppins and Lora typography with fallbacks, and rules for applying them to any artifact the agent produces, with accent colors cycling through shapes and smart color selection against backgrounds.

Installed as-is it makes things look like Anthropic, which is rarely what you want. Its real value is as the reference implementation of a pattern: fork it, swap in your own colors, fonts, and usage rules, and every document, slide, and page your agent generates stays inside your identity. At 65K installs, plenty of teams have taken the hint.

• Official Anthropic skill with their exact brand system
• Applies palette and typography to any artifact type
• The reference pattern for building your own brand skill
• Pairs with brandkit: one invents, this one enforces