Marketing Skills (Corey Haines)
coreyhaines31/marketingskills52 marketing skills for CRO, copywriting, SEO, analytics, paid ads and growth, built around a shared product-marketing context file.
48 skills for SEO, GEO/AEO, paid media across six ad platforms, WordPress, GoHighLevel and GA4/Search Console analytics, backed by live-data workflows.
A large, coherently organized marketing pack with a real safe-by-design principle (read before mutate, explicit approval for spend) and genuine live-data integrations rather than templated advice.
NotFair gives an agent concrete marketing workflows rather than generic advice, organized around live connected data wherever possible: Search Console and GA4 for SEO and analytics, and direct read/write access to Google Ads, Meta Ads, X Ads, LinkedIn Ads, Reddit Ads, TikTok Ads, WordPress and GoHighLevel through a universal NotFair MCP connection (OAuth sign-in in Cursor and Gemini CLI, or codex mcp login NotFair). Individual skills include backlink-audit (works without a paid tool using the Search Console Links report, or a provided Ahrefs/Semrush/Majestic/DataForSEO export), broken-link-checker, competitor-pages (gap analysis against specific competing URLs), geo-optimizer (visibility in ChatGPT, Perplexity, Gemini, Claude and AI Overviews), and a full paid-media family (setup, launch, review, optimize, creative, and platform-specific audit/operate skills for each of the six ad networks plus Amazon Ads and bounded ChatGPT Ads experiments).
Each skill's own SKILL.md is very short (about 1 KB) and simply points to a canonical WORKFLOW.md in a parallel category folder (e.g. seo/backlink-audit/WORKFLOW.md), which carries the real procedure, references, scripts and evals; the repo has 26 eval files across the category folders. The README states a stated design principle: read-only review comes before mutation, and paid-media or CRM changes stay explicit and reviewable rather than applied silently. It installs as a Claude Code plugin, Cursor plugin, Kiro power, or Gemini CLI extension, and is also usable through Codex and Hermes.
48 skill entry points averaging about 1 KB that route into a smaller set of category WORKFLOW.md files with supporting references, scripts and 26 eval files across the repository; the thin entry-point pattern means true depth sits one level down from the skill count.
Counted from the repository file tree. How we score depth.
An SEO and advertising plugin whose live account access goes through a hosted NotFair MCP connector (OAuth) that can change ad accounts, WordPress and CRM data after you approve each change, plus local scripts for Search Console, PageSpeed and CMS work. The Search Console preflight uses the gcloud CLI and may log you in, set a quota project and enable Google APIs in your cloud project. A SessionStart reminder hook is documented but only added if you ask, and the version check can be turned off. Only part of the code was read.
gcloud CLI and mint Application Default Credentials access tokens with it.seo/seo-analysis/scripts/_gcloud.py:22gcloud init, gcloud auth application-default login, set the ADC quota project and enable the Search Console and PageSpeed APIs in your active Google Cloud project; the plugin also keeps state in ~/.toprank (config.yaml, ads change logs, update cache).seo/seo-analysis/scripts/preflight.py:147curl https://sdk.cloud.google.com | bash (Google's own installer); it does not run it.seo/seo-analysis/scripts/preflight.py:62The repository (289 files) on Oct 29, 2026. Not a full review: we read 10 of 58 code files; it relies on code outside the repository: Hosted NotFair MCP server and the ad/analytics/CMS accounts it can operate on; Google Cloud SDK (gcloud) used by the Search Console scripts; Most of the 58 listed code files (analyze_gsc.py, CMS fetch scripts, checker.py, content calendar, tests) were not read in full.
NVIDIA SkillSpector static scan: 208 pattern hits (Privilege Escalation 83, Agent Snooping 77, Dangerous Code Execution 11, Supply Chain 10, Excessive Agency 8, Data Flow 5). The CRITICAL tainted-flow hits are the CMS fetch scripts sending a user-supplied CMS token to that user's own Contentful/Ghost endpoint; credential hits are .env/.codexignore entries and doc text; 'Agent Snooping' hits are CHANGELOG and workflow text mentioning mcp.json and ~/.claude paths; sudo/curl lines are gcloud install documentation; 'Do not judge' and 'without disclaimers' are ad-policy and copy-writing prose.
A review, not a certification. Skills change after we check them — read the files before installing.
MIT — commercial use allowed.
Not really. The stated design keeps paid-media and CRM changes explicit and reviewable rather than automatic, which is a safety choice that works against unattended operation.
52 marketing skills for CRO, copywriting, SEO, analytics, paid ads and growth, built around a shared product-marketing context file.
A free Claude Code plugin with 26 SEO sub-skills and 19 agents for audits, schema, AI-search readiness and optional live-data extensions.
Paid-media audit and operations skill covering 12 ad platforms, with per-platform audit skills, versioned JSON reports and approval-gated account changes.
12 skills that draft LinkedIn posts, comments, replies and profile copy in your voice, read engagement data via Apify, and publish through Publora after you approve.
It is an open-source collection of 48 marketing skills for AI agents covering SEO, GEO/AEO, paid advertising across six platforms, WordPress, GoHighLevel and analytics, each backed by a canonical WORKFLOW.md with live-data integrations.
In Claude Code run /plugin marketplace add nowork-studio/notfair-plugin then /plugin install notfair@nowork-studio. Cursor, Kiro, Gemini CLI, Codex and Hermes each have their own install path listed in the README, generally followed by an OAuth login to the NotFair MCP.
Core SEO and analytics work from Search Console and GA4 for free. Richer backlink metrics (Domain Rating/Authority, full link index) need a paid data source like Ahrefs, Semrush or DataForSEO, which the skill is explicit about needing.
The README states paid-media changes stay explicit and that read-only review comes before any mutation, so ad or budget changes are meant to require your approval rather than running automatically.