Claude Skills for Growth Marketing
A collection of skills, AI agents and Claude scripts to automate paid-media, SEO, Search Console and HubSpot audits, and to ship ad-hoc reports. Built for growth professionals who want to run business marketing with method and system.
google-ads-deep-audit
Systematic 50-point audit of a Google Ads account.
View on GitHub arrow_forwardgoogle-ads-anomaly-diagnostic
Diagnose sudden day-over-day shifts in spend or conversions.
View on GitHub arrow_forwardgoogle-ads-ad-copy-writer
Generate RSAs aligned with the brand and the landing page.
View on GitHub arrow_forwardgoogle-ads-negative-keyword-builder
Mine search terms to build negative-keyword lists.
View on GitHub arrow_forwardgoogle-ads-cannibalization-check
Detect keywords competing across campaigns or ad groups.
View on GitHub arrow_forwardgoogle-ads-dayparting-analysis
Bid recommendations by hour of day and day of week.
View on GitHub arrow_forwardmeta-ads-deep-audit
Systematic 50-point review of a Meta Ads account.
View on GitHub arrow_forwardmeta-ads-dayparting-analysis
Hourly Meta delivery analysis to spot optimal windows.
View on GitHub arrow_forwardmeta-ads-frequency-recommendations
Frequency caps and signals to refresh creatives before fatigue.
View on GitHub arrow_forwardlinkedin-ads-deep-audit
Full LinkedIn Ads account review oriented to B2B.
View on GitHub arrow_forwardads-audit
Quick cross-channel audit to diagnose where the problem is.
View on GitHub arrow_forwardads-budget-review
Pacing and budget reallocation recommendations across channels.
View on GitHub arrow_forwardads-creative-audit
Detect creative fatigue, winners and iteration opportunities.
View on GitHub arrow_forwardads-pacing-monitor
Month-to-date pacing alerts against the spend plan.
View on GitHub arrow_forwardads-strategy-plan
Quarterly paid strategy plan: hypotheses, budget and KPIs.
View on GitHub arrow_forwardab-test-design
Statistically valid A/B test design: hypothesis, sample, lift.
View on GitHub arrow_forwardppc-math
Common PPC calculations: CPA, ROAS, target LTV, target CPMs.
View on GitHub arrow_forwardcompetitor-teardown
Teardown of your direct competitors' ads and landings.
View on GitHub arrow_forwardseo-technical-audit
Technical audit of Core Web Vitals, crawlability and indexation.
View on GitHub arrow_forwardseo-schema-audit
Audit of structured data and rich results.
View on GitHub arrow_forwardseo-images-audit
Image-optimization audit (weight, format, alt, lazy).
View on GitHub arrow_forwardseo-local-audit
Local SEO + Google Business Profile audit.
View on GitHub arrow_forwardseo-ai-search-readiness
Readiness for AI search (ChatGPT, Perplexity, Gemini).
View on GitHub arrow_forwardlanding-page-audit
On-page audit of landing pages for conversion and SEO.
View on GitHub arrow_forwardgsc-weekly-report
Weekly Search Console performance report ready for reporting.
View on GitHub arrow_forwardgsc-indexing-audit
Triage indexation issues by impact priority.
View on GitHub arrow_forwardgsc-cannibalization-check
Query/page cannibalization detection.
View on GitHub arrow_forwardgsc-content-opportunities-page2
Identify page-2 keywords with quick-rise potential.
View on GitHub arrow_forwardhubspot-funnel-audit
Audit of HubSpot funnel stages and lifecycle.
View on GitHub arrow_forwardhubspot-attribution-audit
Multi-touch attribution health check in HubSpot.
View on GitHub arrow_forwardhubspot-ads-reconciliation
Reconcile ad clicks against HubSpot contacts.
View on GitHub arrow_forwardhubspot-linkedin-form-sync
LinkedIn Lead Gen Forms → HubSpot sync audit.
View on GitHub arrow_forwarddashboard-builder
Client dashboard snapshot with cross-cut KPIs.
View on GitHub arrow_forwardmemory-update
Per-client memory maintenance: state files + activity log.
View on GitHub arrow_forwardNo skills in this category yet.
Agents that orchestrate audits
10 sub-agents invokable from a skill or directly. Each one has a specific domain: Google Ads, Meta, HubSpot, SEO, tracking and cookies. They compose into multi-agent flows for multi-channel audits.
audit-google
Specialist sub-agent for systematic Google Ads account audits.
View on GitHub arrow_forwardaudit-meta
Specialist sub-agent for Meta Ads account audits.
View on GitHub arrow_forwardaudit-linkedin
Specialist sub-agent for B2B LinkedIn Ads account audits.
View on GitHub arrow_forwardaudit-budget
Sub-agent that validates pacing and cross-channel budget reallocation.
View on GitHub arrow_forwardaudit-creative
Sub-agent that diagnoses creative fatigue, winners and iteration opportunities.
View on GitHub arrow_forwardaudit-tracking
Specialist sub-agent for tracking, tags and attribution audits.
View on GitHub arrow_forwardaudit-seo
Specialist sub-agent for technical and content SEO audits.
View on GitHub arrow_forwardaudit-hubspot
Specialist sub-agent for HubSpot funnel and attribution audits.
View on GitHub arrow_forwardaudit-compliance
Sub-agent that reviews ad compliance for the EU and Spain.
View on GitHub arrow_forwardbusiness-analyst
Sub-agent that synthesizes outputs into business insight: LTV, CAC, payback, margins.
View on GitHub arrow_forwardHelpers that the skills run
Scripts the SEO skills invoke for external API calls: PageSpeed Insights, CrUX History, page fetcher with SSRF guards and HTML parsers. pip install -r scripts/seo/requirements.txt
pagespeed_check.py
Combined PageSpeed Insights v5 + CrUX check for Core Web Vitals.
View on GitHub arrow_forwardcrux_history.py
Core Web Vitals trend analysis with the CrUX History API.
View on GitHub arrow_forwardfetch_page.py
Page fetcher with SSRF guards for safe SEO audits.
View on GitHub arrow_forwardparse_html.py
HTML parsing helpers for on-page analysis.
View on GitHub arrow_forwardseo_credentials.py
Env-var-based API key reader for SEO tools.
View on GitHub arrow_forwardAbout Claude skills
What is a Claude skill?
A skill is a Markdown text file that Claude loads automatically. When you write something that matches the skill's description, Claude follows it as a procedure: it executes the steps defined in the file, calls the necessary tools and returns the result in the established format. It's like a user manual for Claude that helps it execute tasks better. It also helps save tokens and not pay more than necessary for AI usage.
How is a skill different from a prompt or template?
A prompt lives in the session and is lost when it ends. A skill lives in Claude's "memory" and is applied consistently without having to remember it. Also, a skill can invoke specialist AI agents, call MCP server tools, read reference files (benchmarks or compliance) and delegate parts of the work to sub-agents. It's the difference between a loose note and an executable system.
How do I install these skills in my Claude?
It's explained in the GitHub README. You clone the repository and link the skills/ and agents/ folders to your ~/.claude/ (important: install user-scoped, that way you see them in any project) or to the .claude/ of a specific project (project-scoped mode, you'll only see them for that project). I've included the exact commands for installation in both modes. Once linked and Claude restarted, the skills are available automatically.
Do I need an MCP server to use the skills?
Most skills call tools from a custom MCP server that connects with Google Ads, Meta, LinkedIn Ads, Search Console, GA4 and HubSpot. You have three options: 1) build your own equivalent MCP server, 2) adapt the skills so they call the official APIs directly that you can connect with Claude's default integrations, or 3) use them as prompt templates and feed the data manually. In my case I'm using them with an MCP server that connects with all the tools for which I have skills.
Can you help me set up the complete setup in my account?
Yes. I build the Marketing Engineer infrastructure end-to-end: custom MCP server connected to your stack (Google Ads, Meta, LinkedIn, GSC, HubSpot, GA4 — or whatever tool you want; if it has an API, it can be connected), skills and agents configured for your business, and a daily digest in Slack for your team. If you want to operate growth as a system rather than a series of manual tasks, save time and costs, book a diagnosis at /en/marketing-engineer/.
Ready to operate your growth as a system?
Free 30-minute diagnosis: I'll review your current stack, strategy and method. Tell me your situation and I'll reply within 24h.
- check_circle Tracking and attribution audit
- check_circle Stack diagnosis and MCP opportunities
- check_circle Roadmap for the first 20% of improvement