用 Codex 或 Claude 帮你安装 复制这段 Prompt,粘贴到 Codex、Claude 或其他助手里,让它检查 Skill 页面并帮你完成安装。
直接命令不会经过审查 Prompt;运行前请先检查来源。
npx skills add https://github.com/Aradotso/marketing-skills --skill ai-marketing-claude-code-skills命令会保持在同一行。复制前请横向滚动并检查完整内容。
想先保存到本地?可下载 SkillsMP 当前能够提供的文件。
正在显示 SKILL.md
基于 SOC 职业分类
| name | ai-marketing-claude-code-skills |
| description | Install and use marketing framework skills for Claude Code and other AI coding agents |
| triggers | ["install marketing skills for Claude Code","how do I use AI marketing skills","add marketing frameworks to my agent","setup Claude Code skills for marketing","what marketing skills are available","help me with Claude Code marketing skills","configure AI agent marketing expertise","install positioning skills"] |
Skill by ara.so — Marketing Skills collection.
This project provides pre-packaged marketing expertise as "skills" that AI coding agents (Claude Code, OpenClaw, Cursor, etc.) can load and execute. Instead of prompting from scratch, these skills give your agent structured frameworks for positioning, content strategy, LinkedIn authority, homepage audits, and more.
AI Marketing Skills turn marketing frameworks into executable instructions for AI agents. Each skill contains:
The project includes 23 free skills and 10 premium skills covering strategy, content, research, conversion, and productivity.
git clone https://github.com/BrianRWagner/ai-marketing-claude-code-skills.git
cd ai-marketing-claude-code-skills
bash scripts/install.sh
The installer auto-detects Claude Code, OpenClaw, Cursor, and Windsurf and installs to the correct location.
# Clone the repository
git clone https://github.com/BrianRWagner/ai-marketing-claude-code-skills.git
# Create Claude Code skills directory if it doesn't exist
mkdir -p ~/.claude/skills
# Copy all skills
cp -r ai-marketing-claude-code-skills/* ~/.claude/skills/
mkdir -p ~/.openclaw/skills
cp -r ai-marketing-claude-code-skills/* ~/.openclaw/skills/
OpenClaw uses SKILL-OC.md files (token-efficient versions) when available, falling back to SKILL.md if not present.
# From project root
bash scripts/convert.sh --platform=cursor --output-dir=./.cursor/rules
This converts SKILL.md files to .mdc format with proper frontmatter for Cursor.
bash scripts/convert.sh --platform=windsurf --output-dir=./.windsurf/rules
# Install to all detected platforms
bash scripts/install.sh --all
# Install to specific platform only
bash scripts/install.sh --platform=cursor
# Include premium skills (if you've purchased them)
bash scripts/install.sh --include-pro
# Preview without actually installing
bash scripts/install.sh --dry-run
cd ai-marketing-claude-code-skills
bash scripts/list-skills.sh
Output shows:
Strategy & Positioning:
positioning-basics - Core positioning framework (who, what, why different)ai-discoverability-audit - How your brand appears in AI search (ChatGPT, Perplexity, Claude)marketing-principles - Apply timeless principles from Drucker, Ogilvy, Godin, etc.Content & Authority:
linkedin-authority-builder - Build LinkedIn thought leadership systemcontent-idea-generator - Generate content ideas based on ICP and positioningvoice-extractor - Extract and document authentic writing voice from samplesde-ai-ify - Remove AI jargon, restore human voicesocial-card-gen - Generate platform-specific social posts (Twitter, LinkedIn, Reddit)Research & Intelligence:
last30days - Research any topic across Reddit, X, web (last 30 days)reddit-insights - Semantic search and analysis of Reddit contentyoutube-summarizer - Fetch transcripts and generate structured summarieslinkedin-profile-optimizer - Audit and rewrite LinkedIn profile for visibilityConversion & Sales:
homepage-audit - Quick conversion audit for landing pagescold-outreach-sequence - Build personalized LinkedIn/email sequencescase-study-builder - Turn client wins into formatted case studiestestimonial-collector - Systematically gather testimonialsProductivity:
plan-my-day - Energy-optimized daily planning (circadian + GTD)newsletter-creation-curation - Industry-adaptive B2B newsletter creationgo-mode - Deep work execution modeOnce installed, Claude Code automatically detects and uses skills based on your conversation.
You: "Help me with positioning for my SaaS product"
→ Activates positioning-basics skill
You: "How do I show up in ChatGPT search?"
→ Activates ai-discoverability-audit skill
You: "What should I post on LinkedIn this week?"
→ Activates content-idea-generator skill
All v3.1+ skills support three execution modes:
You: "Quick positioning help for my startup"
→ Quick mode: Fast answers, minimal questions
You: "Standard positioning audit"
→ Standard mode: Balanced depth
You: "Deep positioning analysis for enterprise rebrand"
→ Deep mode: Comprehensive audit, all phases
You: "Use the homepage-audit skill on https://example.com"
You: "Run linkedin-authority-builder in deep mode"
You: "Apply marketing-principles to this decision: [paste context]"
Each skill directory contains:
skill-name/
├── SKILL.md # Full Claude Code version (verbose, all phases)
├── SKILL-OC.md # OpenClaw version (token-efficient, ~200 lines)
├── README.md # Human-readable documentation
└── examples/ # Optional usage examples
---
name: skill-name
description: One-line description of what the skill does
triggers:
- "natural phrase 1"
- "natural phrase 2"
- "natural phrase 3"
---
# Skill Name
## Context
[When to use this skill]
## Execution Modes
- **Quick**: Fast answers, 2-3 questions
- **Standard**: Balanced depth, 5-7 questions
- **Deep**: Comprehensive, all phases
## Framework
[Step-by-step methodology]
## Output Format
[Structured deliverable format]
## Examples
[Real usage examples]
Skills that use external APIs require environment variables:
# For skills that search Reddit
export REDDIT_API_KEY="your_reddit_api_key"
# For skills that use web search
export SERP_API_KEY="your_serp_api_key"
# For YouTube transcript fetching
# No API key needed - uses public endpoints
Add to ~/.bashrc, ~/.zshrc, or use a .env file:
# .env file in project root
REDDIT_API_KEY=your_reddit_api_key
SERP_API_KEY=your_serp_api_key
| Platform | Default Skills Location |
|---|---|
| Claude Code | ~/.claude/skills/ |
| OpenClaw | ~/.openclaw/skills/ |
| Cursor | .cursor/rules/ (project-level) |
| Windsurf | .windsurf/rules/ (project-level) |
| Generic | ./ai-marketing-skills/ |
You: "Should I focus on LinkedIn or Twitter for B2B SaaS?"
Claude (using marketing-principles):
→ Applies first principles from Drucker, Godin
→ Evaluates based on your ICP
→ Provides framework-backed recommendation
You: "I need to build LinkedIn presence"
Step 1: Use linkedin-authority-builder
→ Get positioning, content pillars, posting rhythm
Step 2: Use content-idea-generator
→ Generate 30 days of content ideas
Step 3: Use voice-extractor (on your existing content)
→ Document your authentic voice
Step 4: Use social-card-gen
→ Generate platform-specific variants
You: "Research AI agent trends and create a LinkedIn post"
Step 1: Use last30days
→ Research "AI agents" across Reddit, X, web (last 30 days)
→ Get trends, sentiment, key themes
Step 2: Use content-idea-generator
→ Turn insights into post angles
Step 3: Use de-ai-ify
→ Remove AI jargon from final draft
You: "Audit my homepage and LinkedIn profile"
Step 1: Use homepage-audit
→ Get conversion feedback on landing page
Step 2: Use linkedin-profile-optimizer
→ Rewrite headline, about section
→ AI visibility checklist
Step 3: Use ai-discoverability-audit
→ Check how you appear in ChatGPT, Perplexity, Claude
bash scripts/convert.sh --platform=cursor --output-dir=./.cursor/rules
Creates .mdc files with frontmatter:
---
rule: skill-name
description: Skill description
triggers:
- trigger phrase 1
- trigger phrase 2
---
[SKILL.md content converted]
bash scripts/convert.sh --platform=windsurf --output-dir=./.windsurf/rules
Creates .md files with Windsurf-compatible frontmatter.
bash scripts/convert.sh --all
Outputs to ./converted/cursor/, ./converted/windsurf/, etc.
# Create skill directory
mkdir -p ~/.claude/skills/my-custom-skill
# Create SKILL.md
cat > ~/.claude/skills/my-custom-skill/SKILL.md << 'EOF'
---
name: my-custom-skill
description: What this skill does
triggers:
- "trigger phrase 1"
- "trigger phrase 2"
- "trigger phrase 3"
---
# My Custom Skill
## Context
When to use this skill
## Framework
Step-by-step methodology
## Output Format
What the skill delivers
## Examples
Real usage examples
EOF
Claude Code:
# Verify installation path
ls -la ~/.claude/skills/
# Check permissions
chmod -R 755 ~/.claude/skills/
# Restart Claude Code
OpenClaw:
# Check if SKILL-OC.md exists (preferred) or SKILL.md (fallback)
ls -la ~/.openclaw/skills/*/SKILL*.md
# Verify OpenClaw config
cat ~/.openclaw/config.json
# Check environment variables
echo $REDDIT_API_KEY
echo $SERP_API_KEY
# Reload environment
source ~/.bashrc # or ~/.zshrc
# Test API access
curl -H "Authorization: Bearer $REDDIT_API_KEY" https://oauth.reddit.com/api/v1/me
# Make script executable
chmod +x scripts/convert.sh
chmod +x scripts/install.sh
chmod +x scripts/list-skills.sh
# Run with verbose output
bash -x scripts/install.sh
# Check script dependencies
which jq # Required for some conversions
# Manually specify platform
bash scripts/install.sh --platform=cursor
# Check if platform directories exist
ls -la ~/.claude/
ls -la ~/.openclaw/
ls -la ./.cursor/
Premium skills include structured reasoning phases, self-critique loops, and advanced frameworks.
Available on Gumroad: https://brianrwagner.gumroad.com
| Skill | Price | What It Does |
|---|---|---|
| AI Marketing Bundle | $49 | All 7 premium skills |
| AI Discoverability Audit v2 | $19 | Deep audit of AI search presence |
| Founder Intelligence | $15 | Competitive founder/company research |
| Competitor Intel Brief | $12 | Structured competitor analysis |
| Morning Brief System | $14 | Personalized daily intelligence brief |
| Brand Voice Extractor | $9 | Extract brand voice from content samples |
| AI Employee Onboarding | $9 | Onboard AI agents to your business |
| Brand Positioning Audit | $9 | Deep positioning framework audit |
After purchase, extract to pro/ directory and use --include-pro flag:
# Unzip purchased skills
unzip ai-marketing-pro-skills.zip -d ./pro/
# Install with premium skills
bash scripts/install.sh --include-pro
You: "I'm launching a new AI tool for marketers. Help me with positioning."
Claude (using positioning-basics in standard mode):
1. Who is this for? (ICP definition)
2. What do you do? (Core offering)
3. Why are you different? (Unique positioning)
4. What's the category? (Market context)
5. What's the promise? (Value proposition)
→ Delivers: Structured positioning framework document
You: "Build me a LinkedIn content system for SaaS founders"
Claude (using linkedin-authority-builder):
→ Positioning audit
→ 3-5 content pillars
→ Content formats (frameworks, stories, hot takes)
→ Posting rhythm (3-5x/week)
→ 30-day content calendar
Then use content-idea-generator:
→ 30 specific post ideas across pillars
Then use voice-extractor (on your existing writing):
→ Voice guide for AI-assisted writing
You: "Research what people are saying about Notion vs ClickUp in the last 30 days"
Claude (using last30days):
→ Searches Reddit, X, web
→ Aggregates sentiment and themes
→ Identifies switching triggers
→ Highlights feature gaps
Then use competitor-intel-brief (premium):
→ Structured competitive analysis
→ Feature comparison matrix
→ Positioning gaps
→ Opportunity areas
To contribute new skills or improvements:
SKILL.md (full) and SKILL-OC.md (condensed)README.md for human readersSee existing skills for examples.
ai-marketing-claude-code-skills/
├── scripts/
│ ├── install.sh # Auto-install to detected platforms
│ ├── convert.sh # Convert to platform formats
│ └── list-skills.sh # List all available skills
├── positioning-basics/
│ ├── SKILL.md # Claude Code version
│ ├── SKILL-OC.md # OpenClaw version
│ └── README.md # Documentation
├── ai-discoverability-audit/
│ ├── SKILL.md
│ ├── SKILL-OC.md
│ └── README.md
├── [23 free skills total]
├── pro/ # Premium skills (not included)
│ ├── founder-intelligence/
│ ├── competitor-intel-brief/
│ └── [10 premium skills]
└── README.md # This file
MIT License - Free skills are open source. Premium skills are licensed per purchase.