<div align="center">

<img src="assets/infrasity_logo.avif" alt="Infrasity" height="64" />

Claude Code Skills for SEO, GEO & Developer Marketing

Free, open-source Claude skills that audit your docs, score your AI discoverability, and run developer-marketing workflows — so your product gets found, parsed, and cited by AI systems.

Skills · Install · Commands · Who it's for · FAQ · Website

Works with: Claude Code · OpenAI Codex · OpenClaw · Hermes Agent[^hermes] · Mistral Vibe[^vibe] · Cursor · Aider · Windsurf · Kilo Code · OpenCode · Augment · Antigravity

<p align="center"> <img src="./assets/dev-gtm-claude-skills.png" width="100%" alt="dev-gtm-claude-skills"/> </p>

</div>

dev-gtm-claude-skills is a collection of open-source Claude Code skills for SEO, GEO (Generative Engine Optimization), AI discoverability, and developer marketing. Each skill is a self-contained package — a SKILL.md that tells Claude when and how to use it, optional Python tooling, and a README with full usage docs. Install once, then run from Claude Code, Claude Desktop, or Claude.ai with a plain-language prompt.

These skills are built for developer-focused companies — DevTools, AI-agent platforms, observability, and B2B SaaS — that need their documentation and content to be cited by AI systems like ChatGPT, Claude, Perplexity, and Gemini, not just indexed by Google.

What is GEO? Generative Engine Optimization is the practice of optimizing content so it gets surfaced and cited by AI answer engines. Traditional SEO tools optimize for search crawlers; these skills audit the signals — structured content, llms.txt, internal linking, and documentation completeness — that LLMs use when deciding what to recommend.

What you can do in one prompt

• Audit developer docs for SEO and AI discoverability — 33 checks, scored 0–100
• Score API & SDK documentation quality, endpoint by endpoint
• Check AI-readiness — robots.txt, llms.txt, and llms-full.txt in one pass
• Generate 3-month SEO performance reports vs competitors
• Fix internal linking — find orphan pages and dead-ends, get paste-ready suggestions
• Produce SEO content briefs as formatted .docx outlines

What Are Claude Code Skills & Agent Plugins?

Claude Code skills (also called agent skills or coding-agent plugins) are modular instruction packages that give AI coding agents domain expertise they don't have out of the box. Each skill in this repo includes:

• `SKILL.md` — structured instructions, workflows, and decision frameworks that tell the agent when and how to act
• Python tools — optional stdlib-only CLI scripts for the skills that crawl, score, or render reports
• Reference docs — templates, scoring guides, and checklists the skill loads on demand

Because every skill follows the open SKILL.md standard, it isn't locked to a single product. The same package runs in Claude Code, Claude Desktop, Claude.ai, and any other agent that reads the standard (see Multi-Tool Support).

Skills vs Agents

This repo ships several bundles: the SEO/GEO/docs skills under skills/, the full-funnel marketing skills under marketing-skills/, the blog content engine under writing-skills/, a comprehensive SEO suite under seo-skills/, and blog-production agents under agents/.

Table of contents

• What Are Claude Code Skills & Agent Plugins?
• Multi-Tool Support
• Installation
• Skills
• Marketing skills
• Writing skills
• SEO skills
• Notion skills
• Commands
• Who it's for
• Requirements
• Sample outputs
• Repository structure
• FAQ
• Contributing
• License

Multi-Tool Support

Every skill here is written to the open SKILL.md standard, so it isn't tied to a single agent. Claude Code, Claude Desktop, and Claude.ai read the skills natively (see Installation). For every other tool, the repo ships converter and installer scripts under scripts/ — clone the repo, then run the one-liner for your tool.

git clone https://github.com/infrasity-labs/dev-gtm-claude-skills.git
cd dev-gtm-claude-skills

Each script discovers all skills under skills/, marketing-skills/, writing-skills/, notion-skills/, and seo-skills/ and installs them in the format that tool expects.

See full install steps per tool below.

Installation

Most skills need no API keys; the SEO skills that pull live search data use a DataForSEO MCP server (setup below).

Quick install (recommended)

Install every skill in this repo with one command — no cloning, no copying:

npx skills add Infrasity-Labs/dev-gtm-claude-skills

This pulls the latest skills straight into your Claude Code skills directory. Re-run it any time to update. Prefer to install manually or use Claude Desktop / Claude.ai? Use one of the methods below.

Blog skills runtime (agents + scripts)

The writing skills add a /blog content engine built from 30 sub-skills, 5 subagents, and shared Python scripts. npx skills add installs the skill instructions, but the subagents and shared scripts need one extra step so the /blog write pipeline (research → write → SEO → review) runs end-to-end:

# Clone the repository to access the installer
git clone https://github.com/Infrasity-Labs/dev-gtm-claude-skills.git
cd dev-gtm-claude-skills
./scripts/claude-blog-install.sh                  # agents → ~/.claude/agents, scripts → ~/.claude/scripts
pip install -r writing-skills/requirements.txt    # textstat + beautifulsoup4

Use --project to install into the current project's ./.claude/ instead of ~/.claude/ (make sure to run the script from your project's root directory, e.g., path/to/cloned-repo/scripts/claude-blog-install.sh --project), and --dry-run to preview. A few sub-skills need their own credentials — blog-google (Google API OAuth), blog-audio / blog-image (GOOGLE_AI_API_KEY + nanobanana MCP), blog-notebooklm (browser login) — and all degrade gracefully when unconfigured.

Claude Code (manual)

Clone the repo and copy the skill into your Claude Code skills directory.

git clone https://github.com/infrasity-labs/dev-gtm-claude-skills.git

Project-level (available only in the current project):

mkdir -p .claude/skills
cp -r dev-gtm-claude-skills/skills/<skill-name> .claude/skills/

User-level (available across all projects):

mkdir -p ~/.claude/skills
cp -r dev-gtm-claude-skills/skills/<skill-name> ~/.claude/skills/

Skills activate automatically — Claude reads every SKILL.md in .claude/skills/ at the start of each session. Trigger them by describing the task in plain language, or type /<skill-name> directly.

<p align="center"> <img src="./assets/clone-repo.png" width="100%" alt="clone repo"/> </p>

<p align="center"> <img src="./assets/activate-skills.gif" width="100%" alt="Activate skills GIF"/> </p>

Claude Desktop / Claude.ai

Skills upload as ZIP files via Settings → Customize → Skills → Create skill → Upload a skill.

git clone https://github.com/infrasity-labs/dev-gtm-claude-skills.git
cd dev-gtm-claude-skills/skills

# Zip the skill you want to install
zip -r docs-auditor.zip docs-auditor/

Upload the .zip, toggle the skill on, and it's active across all your chats. Uploaded skills stay private to your account; Claude installs any required packages the first time it runs them.

<p align="center"> <img src="./assets/converting-to-zip.gif" width="100%" alt="Converting to zip GIF"/> </p>

<p align="center"> <img src="./assets/add-to-claude.gif" width="100%" alt="Add to Claude GIF"/> </p>

DataForSEO MCP (for the live-search skills)

growth-report, blog-post-counter, and api-docs-quality-report pull live search data via a DataForSEO MCP server.

Claude Code — add to .claude/settings.json or ~/.claude/settings.json:

{
  "mcpServers": {
    "dataforseo": {
      "command": "npx",
      "args": ["-y", "@dataforseo/mcp-server"],
      "env": {
        "DATAFORSEO_USERNAME": "[email protected]",
        "DATAFORSEO_PASSWORD": "your_api_password"
      }
    }
  }
}

Claude Desktop — Customize → Connectors → Add Custom Connector:

https://your_email:[email protected]/http

Get credentials at dataforseo.com.

Clay connector (for prospect enrichment and cold-email)

dev-marketing-prospector and cold-email use Clay's data enrichment tools to look up company funding, headcount, tech stack, and verified contact information — replacing manual research across Crunchbase, LinkedIn, and Apollo.

Claude.ai / Claude Desktop — Settings → Customize → Connectors → Connect Clay. Sign in with your Clay account. Once connected, the find-and-enrich-company, ask-question-about-accounts, and find-and-enrich-contacts-at-company tools become available automatically.

Claude Code — add to .claude/settings.json or ~/.claude/settings.json:

{
  "mcpServers": {
    "clay": {
      "command": "npx",
      "args": ["-y", "@clay-hq/mcp-server"],
      "env": {
        "CLAY_API_KEY": "your_clay_api_key"
      }
    }
  }
}

Get your API key at clay.com. Both skills degrade gracefully when Clay is not connected — dev-marketing-prospector falls back to manual web search across all source categories, and cold-email prompts the user for research signals instead.

Notion connector (for saving outputs to your workspace)

competitor-profiling, customer-research, and content-brief save their outputs to Notion as structured databases — competitor profiles become queryable competitive intelligence records, VOC research becomes a searchable quote bank, and content briefs sync to a living Notion database your whole team can filter and reference.

Claude.ai / Claude Desktop — Settings → Customize → Connectors → Connect Notion. Sign in with your Notion account and grant access to the workspaces you want skills to write to. Once connected, the notion-search, notion-create-database, notion-create-pages, and notion-update-page tools become available automatically.

Claude Code — add to .claude/settings.json or ~/.claude/settings.json:

{
  "mcpServers": {
    "notion": {
      "command": "npx",
      "args": ["-y", "@notionhq/notion-mcp-server"],
      "env": {
        "NOTION_API_KEY": "your_notion_integration_secret"
      }
    }
  }
}

Get your integration secret at notion.so/my-integrations. All skills degrade gracefully when Notion is not connected — outputs are returned in chat or saved locally, and each skill appends a 💡 note with setup instructions.

Apify connector (for web scraping — reviews, Reddit, Google Maps)

reddit-comments, competitor-profiling, customer-research, seo-maps, and programmatic-seo use Apify actors to scrape data that WebFetch and Firecrawl can't reliably reach — Reddit threads blocked by rate limits, G2/Capterra/Trustpilot review pages with anti-bot protection, and live Google Maps business listings.

Note: Apify is an official Anthropic connector and is available on Claude Desktop only. It cannot be configured as a Claude Code MCP server.

Claude Desktop — Settings → Customize → Connectors → Connect Apify. Sign in with your Apify account. Once connected, the following actors are available across skills:

All skills degrade gracefully when Apify is not connected — each falls back to WebFetch or Firecrawl and appends a 💡 note explaining what Apify would unlock.

Klaviyo connector (for email template and campaign sync)

emails, churn-prevention, lead-magnets, and onboarding use Klaviyo's MCP tools to read your existing email setup and push new templates directly into your account — auditing existing flows, creating HTML and drag-and-drop templates, building campaigns, and pulling open/click/revenue metrics per flow step.

Note: The Klaviyo MCP can create templates and campaigns but cannot create automated flows. After the skill builds your templates, configure trigger conditions, delays, and branching logic in the Klaviyo UI.

Claude.ai / Claude Desktop — Settings → Customize → Connectors → Connect Klaviyo. Sign in with your Klaviyo account. Once connected, tools like list_email_templates, create_email_template, get_flows, create_campaign, and query_metric_aggregates become available automatically.

Claude Code — add to .claude/settings.json or ~/.claude/settings.json:

{
  "mcpServers": {
    "klaviyo": {
      "command": "npx",
      "args": ["-y", "@klaviyo/mcp-server"],
      "env": {
        "KLAVIYO_API_KEY": "your_klaviyo_private_api_key"
      }
    }
  }
}

Get your private API key at klaviyo.com/account#api-keys-tab. All skills degrade gracefully when Klaviyo is not connected — they output email copy and sequence structure in chat, which you can manually import into any ESP.

Install Steps for Every Tool

All CLI installs start with cloning the repo:

git clone https://github.com/infrasity-labs/dev-gtm-claude-skills.git
cd dev-gtm-claude-skills

Hermes Agent ✅

1. Run the sync script:

   python3 scripts/sync-hermes-skills.py

1. Skills are symlinked into ~/.hermes/skills/. Add --copy if you want to delete the repo clone after install.
2. Open Hermes and describe your task — the agent auto-loads the matching skill. You can also invoke by name: /docs-auditor.

Pass --verbose to see each skill as it syncs, --dry-run to preview without installing.

Kilo Code ✅

1. In VS Code settings, enable the compatibility flag:

   kilo-code.new.claudeCodeCompat: true

(Search "kilo-code claudeCodeCompat" in the VS Code settings UI.)

1. Install skills using the standard Claude Code method — either the quick install:

   npx skills add Infrasity-Labs/dev-gtm-claude-skills

or manual copy to ~/.claude/skills/:

   mkdir -p ~/.claude/skills
   cp -r skills/<skill-name> ~/.claude/skills/

1. Open a Kilo Code chat and describe your task. Kilo Code reads from ~/.claude/skills/ when claudeCodeCompat is on.

Note: Placing .md files in .kilocode/rules/ only injects passive context — it does not enable skill invocation. The claudeCodeCompat path is the correct approach.

Mistral Vibe ✅

1. Run the install script:

   ./scripts/vibe-install.sh

1. Skills are symlinked flat into ~/.vibe/skills/<skill-name>/. Add --copy to copy instead of symlink.
2. Open Vibe and describe your task, or invoke directly: /docs-auditor. Vibe auto-discovers skills from ~/.vibe/skills/.

Pass --dry-run to preview without installing, --verbose to list each skill.

OpenAI Codex ✅

1. Generate local symlinks:

   python3 scripts/sync-codex-skills.py

1. Copy skills to the Codex global directory:

   ./scripts/codex-install.sh

1. Skills land in ~/.codex/skills/<skill-name>/. Open Codex and describe your task — the model reads the skill automatically.

Use ./scripts/codex-install.sh --dry-run to preview, --list to see all available skills.

OpenClaw ✅

1. Run the install script:

   ./scripts/openclaw-install.sh

1. Restart the OpenClaw gateway to pick up the new skills:

   openclaw gateway restart

1. Verify the skills are visible:

   openclaw skills list

1. Open an OpenClaw chat and describe your task. The agent auto-loads the matching skill. You can also invoke explicitly: /docs-auditor.

Use --dry-run to preview without installing, --force to overwrite existing skills.

Augment ✅

Augment reads skills from the .augment/skills/ directory inside your project — run these commands from your project root, not from the cloned repo.

1. Convert skills to Augment format:

   cd /path/to/dev-gtm-claude-skills
   ./scripts/convert.sh --tool augment

1. Install into your project:

   ./scripts/install.sh --tool augment --target /path/to/your/project

Skills land in /path/to/your/project/.augment/skills/<skill-name>/SKILL.md.

1. Open Augment in that project and describe your task — Augment detects the matching skill from .augment/skills/ automatically.

Re-run both commands whenever you want to update to the latest skills. Use --force on install.sh to overwrite.

Antigravity ✅

1. Convert skills to Antigravity format:

   ./scripts/convert.sh --tool antigravity

1. Install globally:

   ./scripts/install.sh --tool antigravity

1. Skills land in ~/.gemini/antigravity/skills/<skill-name>/SKILL.md. Open Antigravity and describe your task — the agent auto-loads the matching skill.

Re-run both commands to update. Use --force on install.sh to overwrite existing skills.

Cursor 🔜 (steps ready, not yet tested)

1. Convert skills to Cursor .mdc rule format:

   ./scripts/convert.sh --tool cursor

1. Install into your project:

   ./scripts/install.sh --tool cursor --target /path/to/your/project

Skills land in /path/to/your/project/.cursor/rules/<skill-name>.mdc.

1. Open Cursor in that project — rules are applied automatically per the globs and alwaysApply settings in each .mdc file.

Aider 🔜 (steps ready, not yet tested)

1. Convert all skills into a single CONVENTIONS.md:

   ./scripts/convert.sh --tool aider

1. Install into your project:

   ./scripts/install.sh --tool aider --target /path/to/your/project

1. Run Aider with the conventions file:

   aider --read CONVENTIONS.md

Windsurf 🔜 (steps ready, not yet tested)

1. Convert skills to Windsurf skill bundle format:

   ./scripts/convert.sh --tool windsurf

1. Install into your project:

   ./scripts/install.sh --tool windsurf --target /path/to/your/project

Skills land in /path/to/your/project/.windsurf/skills/<skill-name>/SKILL.md.

1. Open Windsurf in that project and describe your task.

Skills

Ten production-grade skills for SEO, GEO, documentation, and developer-marketing workflows.

Marketing skills

A full-funnel pack of 28 developer-marketing skills under marketing-skills/ — covering demand gen, content, lifecycle, conversion, and retention. Install the whole set with the quick install command above.

Writing skills

A full blog content engine under writing-skills/ — 30 sub-skills routed by a blog orchestrator, plus 5 production subagents (agents/) and shared Python tooling (scripts/). Dual-optimized for Google rankings (E-E-A-T, Dec 2025 Core Update) and AI citations (GEO/AEO). Invoke everything through /blog <subcommand> — see the /blog command table. The same folder also ships 4 standalone writing skills for copy, social, and prose quality.

These need one extra setup step beyond npx skills add — see Blog skills runtime.

Standalone writing skills

Beyond the /blog engine, writing-skills/ ships 4 standalone craft skills for copy, social, and prose quality — usable on their own, no runtime setup required.

SEO skills

A comprehensive SEO suite under seo-skills/ — 25 sub-skills routed by a seo orchestrator, plus 18 specialist subagents (agents/), shared Python tooling (scripts/), and 8 optional MCP extensions (seo-skills/extensions/). Dual-optimized for Google rankings (technical SEO, E-E-A-T, Core Web Vitals with INP) and AI visibility (GEO for AI Overviews/ChatGPT/Perplexity). Detects business type (SaaS, e-commerce, local, publisher, agency) and delegates to the right specialists. Invoke everything through /seo <subcommand> — see the /seo command table.

Standalone SEO skill

Alongside the suite, seo-skills/ ships `geo-seo-claude` — a self-contained SEO/GEO/AEO optimizer (audit, schema generation, metadata validation, keyword/entity extraction, IndexNow) with its own bundled scripts and reference library.

SEO extensions

8 optional MCP extensions under seo-skills/extensions/ add live data and AI-visibility coverage — each installed on demand via its own install.sh:

Web design skills

Four skills for visual design, UI auditing, and site planning. Install via the standard Claude Code or CLI method — they live in web-design/.

Product management skills

Eight skills for PM workflows — strategy, PRDs, prioritization, backlog, and growth. Install via the standard Claude Code or CLI method — they live in product-management-skills/.

Notion skills

Seven personal-productivity and workflow skills under notion-skills/ — brain dumps, email triage, meeting notes, SEO content briefs, CLAUDE.md generation, mid-conversation reassessment, and multi-agent workflow scripting. Install via the standard Claude Code or CLI method.

Commands

If you've installed the command pack, every skill is also a slash subcommand. The SEO, GEO, and docs skills run under /dev-gtm; the full-funnel marketing skills run under /marketing; the comprehensive SEO suite runs under /seo; the blog content engine runs under /blog; the web design skills run under /web-design; the product management skills run under /pm; and the Notion and productivity skills run under /notion.

<p align="center"> <img src="./assets/claude-commands.png" alt="Claude Commands" width="400"/> </p>

/dev-gtm — SEO, GEO & docs

/marketing — full-funnel growth

/blog — blog content engine

Single entry point for all 30 writing skills. Requires the Blog skills runtime step.

/seo — comprehensive SEO suite

Single entry point for the SEO suite. The seo orchestrator detects business type and delegates to specialist subagents in parallel for full audits.

/web-design — design & UI

/pm — product management

/notion — productivity & workflow

Who it's for

These skills are built for technical teams who care about being discoverable by AI, not just by Google:

• DevTool & infrastructure startups measuring whether their docs are LLM-citable
• AI-agent and LLM platforms that need their API/SDK docs parsed correctly by agents
• Developer marketing & DevRel teams running SEO + GEO audits without a dedicated SEO hire
• Technical content teams producing briefs and benchmarking content against competitors
• Founders who want a fast, honest read on their AI visibility before investing in it

Requirements

• Claude Code, Claude Desktop, or Claude.ai
• Python 3.10+
• Node.js with npx (to run the DataForSEO MCP server locally)
• A DataForSEO API account — only for the live-search skills: growth-report, blog-post-counter, api-docs-quality-report

Sample outputs

Repository structure

dev-gtm-claude-skills/
├── assets/                          # Logos, images, and GIFs for this README
├── skills/
│   ├── docs-auditor/
│   │   ├── SKILL.md                 # Agent instructions (Claude reads this)
│   │   ├── README.md                # Human-facing usage docs
│   │   └── references/              # Scoring guide, widget template, fetch strategy
│   ├── sdk-docs-auditor/
│   ├── api-docs-quality-report/
│   ├── growth-report/
│   ├── blog-post-counter/
│   ├── brief-outline-generator/
│   ├── llms-txt-checker/
│   ├── orphan-pages-internal-linking-opportunities/
│   ├── no-outlinks-audit/
│   └── dev-marketing-prospector/
├── marketing-skills/                # 28 full-funnel developer-marketing skills
│   ├── seo-audit/
│   ├── content-strategy/
│   ├── cro/
│   ├── emails/
│   └── …                            # ad-creative, ads, pricing, launch, and more
├── notion-skills/                   # 7 personal-productivity and workflow skills (/notion)
│   ├── capture/                     # brain-dump organizer
│   ├── inbox/                       # email triage system
│   ├── reflect/                     # mid-conversation reassessment
│   ├── workflow-builder/            # multi-agent workflow script generator
│   ├── content-brief/               # keyword → SEO brief → Notion
│   ├── claude-md-starter/           # repo scanner + CLAUDE.md generator
│   └── meeting-notes/               # structured meeting summaries
├── writing-skills/                  # 30 blog/content-engine sub-skills (/blog) + 4 standalone
│   ├── blog/                        # orchestrator
│   ├── blog-write/
│   ├── blog-analyze/
│   ├── ogilvy-copywriting/          # standalone: sales-driven copy
│   ├── social-media-posts/          # standalone: per-platform social posts
│   ├── reddit-comments/             # standalone: on-brand Reddit replies
│   ├── stop-slop/                   # standalone: strip AI writing tells
│   └── …                            # rewrite, brief, schema, translate, and more
├── seo-skills/                      # 25 SEO sub-skills (/seo) + geo-seo-claude + 8 extensions
│   ├── seo/                         # orchestrator
│   ├── seo-audit/                   # full-site audit (installs as seo-audit-full)
│   ├── seo-technical/
│   ├── seo-geo/
│   ├── geo-seo-claude/              # standalone SEO/GEO/AEO optimizer (own scripts)
│   ├── extensions/                  # 8 optional MCP extensions (dataforseo, firecrawl, …)
│   └── …                            # schema, local, maps, backlinks, cluster, and more
├── agents/                          # blog + 18 SEO specialist subagents (Task subagents)
├── data/                            # shared data (e.g., google-updates.json)
├── scripts/                         # shared blog + SEO scripts + multi-tool installers
└── README.md

FAQ

Are these Claude skills free? Yes, every skill is free and MIT-licensed. No license fees, no paywalls.

Do I need Claude Code? No. Skills run in Claude Code, Claude Desktop, or Claude.ai. Desktop and web use the ZIP upload method above.

Which skills need API keys? Most need none. Only the live-search skills (growth-report, blog-post-counter, api-docs-quality-report) use DataForSEO, and the internal-linking skills (orphan-pages…, no-outlinks-audit) use Ahrefs.

How is this different from a standard SEO audit tool? Traditional tools optimize for Google crawlers. These skills are built for developer documentation and AI discoverability. They check whether your docs can be parsed, cited, and recommended by LLMs.

How often do new skills ship? Regularly. Star or watch the repo to get notified.

Contributing

To add a skill, create a folder under skills/:

skills/your-skill-name/
├── SKILL.md          # Required: agent instructions
├── README.md         # Required: human-facing docs
├── requirements.txt  # If the skill has Python dependencies
└── references/       # Templates, scoring guides, or other reference files

SKILL.md must include a frontmatter block:

---
name: your-skill-name
description: >
  One paragraph describing exactly when Claude should activate this skill.
  Write it to match the natural language a user would use.
---

The description field is what Claude uses to decide when to activate the skill — be specific about trigger phrases and input formats. Open a PR with your skill folder and a one-line summary.

License

MIT — see LICENSE.

Built by Infrasity

Infrasity is a developer-marketing agency for DevTools, AI-agent platforms, and B2B SaaS. We help technical products get found, parsed, and cited by AI systems.

🌐 infrasity.com · 🔗 LinkedIn · 📸 Instagram

Want a GEO audit of your own docs? → infrasity.com/claude-skills