Skip to main content

tech-marketing-framework

Turn Claude Code or Codex into a senior product marketer with repeatable skills for developer tools content, messaging, and campaign generation

ソース情報

リポジトリ
reason-machines/marketing-skills
ソースの最終更新活動
2026年5月18日 22:26
検出された SKILL.md の言語
英語
スター
10
フォーク
1

インストール方法

デフォルトでは、最初にソースを確認する Prompt が選択されています。直接コマンドに切り替えるか、ローカルコピーをダウンロードすることもできます。

ソースファイルを確認

インストールを決める前に、SKILL.md と SkillsMP に表示されている付属ファイルをお読みください。

SKILL.md を表示中

SKILL.md
ソースの指示 · 読み取り専用プレビュー
name
tech-marketing-framework
description
Turn Claude Code or Codex into a senior product marketer with repeatable skills for developer tools content, messaging, and campaign generation
triggers
["help me create marketing content for my developer tool","generate a messaging framework for my technical product","build a product marketing strategy","create social posts for my SaaS launch","write a sales deck for my API product","set up marketing automation for my dev tool","run a messaging workshop","generate product hunt launch assets"]
# Tech Marketing Framework > Skill by [ara.so](https://ara.so) — Marketing Skills collection. A comprehensive marketing framework that turns Claude Code and Codex into a senior product marketer for technical audiences. Provides identity guidelines, content standards, template docs, and autonomous skills for content generation, messaging workshops, and campaign optimization. ## What It Does This project provides: - **Identity and voice guidelines** that prioritize technical accuracy over marketing fluff - **Content guidelines** with strict standards for writing, formatting, and AI discoverability - **Template docs** for product briefs, personas, competitive intel, and messaging - **Skills for content generation** (email sequences, social posts, messaging workshops) - **Autoresearch** for autonomous skill optimization via binary evals ## Installation **Prerequisites:** [Claude Code CLI](https://docs.anthropic.com/en/docs/claude-code) or [OpenAI Codex CLI](https://help.openai.com/en/articles/11096431-openai-codex-ci-getting-started) ```bash # Install Claude Code CLI npm install -g @anthropic-ai/claude-code # Or install OpenAI Codex CLI npm install -g @openai/codex # Clone the framework into your project git clone https://github.com/j1ngg/tech-marketing-framework.git cd tech-marketing-framework # Run Claude Code claude # Or run Codex codex ``` Claude Code automatically loads `claude.md` and `.claude/` directory. Codex automatically loads `AGENTS.md` and `.agents/skills/` directory. ## Project Structure ``` ├── AGENTS.md # Codex repo instructions ├── claude.md # Claude Code instructions ├── .agents/skills/ # Codex skill wrappers ├── .claude/ │ ├── rules/ │ │ └── content-guidelines.md # Writing standards │ ├── skills/ # Marketing skill workflows │ │ ├── messaging-positioning/ │ │ ├── social-posts/ │ │ ├── email/ │ │ ├── ads/ │ │ ├── sales-deck/ │ │ ├── blog/ │ │ ├── editorial-calendar/ │ │ ├── image/ │ │ ├── launch-roundup/ │ │ ├── producthunt-launch/ │ │ └── autoresearch/ │ └── agents/ │ ├── asset-reviewer.md │ ├── ads-auditor.md │ └── how-they-market.md ├── docs/ │ ├── inputs/ # Product-specific templates │ │ ├── product_brief.md │ │ ├── target_personas.md │ │ ├── messaging_positioning.md │ │ ├── competitor_intel.md │ │ ├── testimonials.md │ │ └── brand_guidelines.md │ └── reference/ │ └── channel_directory.md └── output/ # Generated assets ├── social/ ├── blog/ ├── email/ ├── ads/ ├── decks/ ├── images/ └── research/ ``` ## Getting Started ### 1. Fill in Product Templates Navigate to `docs/inputs/` and complete these foundational documents: ```bash # Edit product brief with your product details vi docs/inputs/product_brief.md # Define your target personas vi docs/inputs/target_personas.md # Add competitor intelligence vi docs/inputs/competitor_intel.md # Include approved customer testimonials vi docs/inputs/testimonials.md ``` **Required fields in product_brief.md:** - Product name and tagline - Core problem solved - Key features and capabilities - Target audience - Unique value proposition **Required fields in target_personas.md:** - Persona name and role - Pain points - Goals and success metrics - Preferred channels - Technical sophistication level ### 2. Run Your First Skill In Claude Code, use slash commands: ``` /messaging-positioning ``` In Codex, use dollar sign or plain English: ``` $messaging-positioning ``` or simply ask: ``` Help me build a messaging framework for my product ``` ## Core Skills ### Messaging & Positioning Workshop Interactive workshop that builds your messaging foundation: ``` /messaging-positioning ``` Walks through four discovery sections: 1. **Vision** - Long-term company direction 2. **Product** - Capabilities and differentiation 3. **Competitive** - Market alternatives 4. **Targeting** - ICP and personas **Output:** `docs/inputs/messaging_positioning.md` ### Social Media Posts Generate platform-optimized social content: ``` /social-posts ``` **Supported platforms:** - LinkedIn (brand and personal accounts) - Twitter/X (threads and standalone posts) **Process:** 1. Select platform and account type 2. Provide source material (blog post, feature announcement, etc.) 3. Skill generates 3-5 variants optimized for each platform's audience **Output:** `output/social/[platform]-[timestamp].md` ### Email Sequences Generate programmatic email sequences: ``` /email ``` **Current support:** Event follow-up sequences (post-conference, post-meeting, post-dinner) **4-email sequence structure:** | Email | Timing | Purpose | |-------|--------|---------| | 1 | Day 0 | Event recall + pain points (no link) | | 2 | +2 days | Peer social proof + product demo | | 3 | +6 days | Customer results + resource link | | 4 | +11 days | Soft close + problem list (no link) | **Key rules:** - Real links only (no placeholders) - Approved customers only (reads from `testimonials.md`) - Programmatic personalization (shared context, not fabricated 1:1 details) **Output:** `output/email/[sequence-type]-[timestamp].md` ### Paid Ads Generator Generate ad copy for Google, Meta, and LinkedIn: ``` /ads ``` Uses creative matrix: Topics × Personas × Angles **Supported platforms:** - Google Ads (Search, Display) - Meta (Facebook, Instagram) - LinkedIn Ads **Each variant includes:** - Headlines (platform character limits enforced) - Descriptions - Image/creative briefs - Target audience mapping **Output:** `output/ads/[platform]-[timestamp].md` ### Sales Deck Builder Create B2B sales narrative decks using April Dunford's methodology: ``` /sales-deck ``` **8-step structure:** **Phase 1: The Setup (Slides 1-10)** - No product mentioned 1. Title/Thesis - Bold provocative insight 2. TLDR - Executive summary 3-4. Status Quo - Why current approach was rational 5-6. Disruption - What changed 7-8. Alternatives - Why other options fall short 9-10. Perfect World - New purchase criteria **Phase 2: Follow-Through (Slides 11-17)** 11. Introduction - Product positioning 12-14. Differentiated Value - Capabilities mapped to criteria 15. Proof - Customer results 16. Objections (optional) 17. The Ask - CTA **Output:** - Markdown outline: `output/decks/[product-name]-sales-deck.md` - PPTX file: `output/decks/[product-name]-sales-deck.pptx` ### Blog Post Generator Generate SEO and AEO optimized blog posts: ``` /blog ``` **Supported blog types:** | Type | Best For | Length | |------|----------|--------| | Feature announcement | Release notes, new capability | 800-1,200 words | | Product launch | New product, major release | 1,200-1,800 words | | Thought leadership | Opinion piece, contrarian take | 1,500-2,500 words | | Cookbook/tutorial | How-to guide, integration walkthrough | 1,500-3,000 words | | Report conversion | Whitepaper/case study → blog | 1,000-1,500 words | **Every blog includes:** - TLDR section (LLM-extractable) - Metadescription (<160 chars) - FAQ section (3-5 questions, AEO optimized) - Schema markup suggestions - Internal linking recommendations - Visual callout flags **Tri-publish variants:** - Website/blog (long-tail keywords, schema) - LinkedIn Pulse (DR 98 domain, LLM citations) - X article (For You feed optimization) **Output:** `output/blog/[title-slug]-[timestamp].md` ### Editorial Calendar Build a monthly rolling content plan using MKT1 methodology: ``` /editorial-calendar ``` **Workshop sections:** 1. Business Context - Goals, revenue levers, GTM motion 2. Perceptions - What audience should say about you 3. Content Pillars & Audience - Topics, channels, formats 4. Capacity - Team size, cadence, backlog 5. Product & Market Context - Launches, events, competitive moves **Output artifacts:** - 3-Month Theme Roadmap - 4-Week Editorial Calendar - Monthly Refresh Checklist **Output:** `output/editorial/[month]-editorial-calendar.md` ### Product Hunt Launch Generate complete Product Hunt launch package: ``` /producthunt-launch ``` **Generated assets:** - Product name and tagline - Description (260 chars) - Gallery images brief (3 required) - First comment (detailed walkthrough) - Hunter outreach template - 8-week launch timeline with milestones **Output:** `output/launch/producthunt-[product-name].md` ## Agents ### Asset Reviewer Reviews generated assets against content guidelines: ``` /asset-reviewer ``` Point it at any generated asset in `output/` to get compliance scoring. ### Ads Auditor Audits ad performance with health scoring: ``` /ads-auditor ``` Requires ad performance data (CSV export from ad platform). ### Competitive Intelligence Analyzes how others market similar products: ``` /how-they-market ``` **Input:** Competitor URL or product name **Output:** `output/research/competitor-[name]-analysis.md` ## Configuration ### Environment Variables If using image generation skill (requires MCP): ```bash # .env ANTHROPIC_API_KEY=your_api_key_here ``` ### Brand Guidelines Customize voice and style in `docs/inputs/brand_guidelines.md`: ```markdown ## Voice Principles - Technical accuracy over marketing fluff - Show, don't tell (code examples, not promises) - Developers are sophisticated buyers ## Writing Rules - Use active voice - Lead with the insight - No empty superlatives ``` ## Content Guidelines The framework enforces strict content standards defined in `.claude/rules/content-guidelines.md`: **Core principles:** - Technical accuracy is non-negotiable - No marketing fluff or empty superlatives - Code examples must be real and runnable - Claims require proof (links, customer quotes, data) **Formatting standards:** - Markdown-first - Archetypal phrasing for LLM extraction - Answer-first structure - Schema markup where applicable ## Common Workflows ### Launch a New Feature ``` # 1. Update product brief with new feature vi docs/inputs/product_brief.md # 2. Generate feature announcement blog /blog # 3. Create social promotion /social-posts # 4. Build email sequence /email # 5. Update sales deck /sales-deck ``` ### Quarterly Marketing Refresh ``` # 1. Run messaging workshop /messaging-positioning # 2. Build editorial calendar /editorial-calendar # 3. Generate ad variants /ads # 4. Review all assets /asset-reviewer ``` ### Competitive Response ``` # 1. Analyze competitor marketing /how-they-market # 2. Update competitive intel vi docs/inputs/competitor_intel.md # 3. Refresh messaging /messaging-positioning # 4. Generate new ad variants /ads ``` ## Advanced Features ### Autoresearch (Skill Optimization) The framework includes a meta-skill that autonomously optimizes other skills: ``` /autoresearch ``` **How it works:** 1. Runs target skill multiple times with variations 2. Applies binary evals to outputs 3. Updates skill definition based on what works 4. Commits improvements **Output:** `output/research/autoresearch-[skill-name]-[timestamp].md` ### Skill Builder Create new custom skills: ``` /skill-builder ``` Walks through skill creation workflow and generates new skill definition in `.claude/skills/`. ## Troubleshooting ### Skills Not Loading **Claude Code:** ```bash # Verify claude.md exists ls -la claude.md # Verify .claude directory structure ls -la .claude/skills/ ``` **Codex:** ```bash # Verify AGENTS.md exists ls -la AGENTS.md # Verify .agents directory structure ls -la .agents/skills/ ``` ### Missing Product Context Skills pull from `docs/inputs/`. If getting generic output: ```bash # Check which templates are filled in ls -la docs/inputs/ # Minimum required: product_brief.md and target_personas.md ``` ### Generated Content Too Generic Ensure you've completed: 1. `messaging_positioning.md` - Provides voice and positioning 2. `testimonials.md` - Provides proof points 3. `competitor_intel.md` - Provides differentiation context ### Image Generation Fails Image skill requires MCP (Model Context Protocol) setup: ```bash # Verify MCP is configured claude mcp status # Check API key is set echo $ANTHROPIC_API_KEY ``` ## Best Practices 1. **Start with messaging workshop** - Run `/messaging-positioning` before generating assets 2. **Keep inputs updated** - Treat `docs/inputs/` as source of truth 3. **Review generated assets** - Always run `/asset-reviewer` before publishing 4. **Maintain testimonials** - Add new customer quotes to `testimonials.md` regularly 5. **Refresh competitive intel** - Run `/how-they-market` quarterly 6. **Version control outputs** - Commit generated assets for team review 7. **Use tri-publish strategy** - Maximize content ROI with blog → LinkedIn Pulse → X ## File Locations All generated assets go to `output/` directory: ``` output/ ├── social/ # Social media posts ├── blog/ # Blog posts and articles ├── email/ # Email sequences ├── ads/ # Ad copy and creative briefs ├── decks/ # Sales and pitch decks ├── images/ # Generated marketing images └── research/ # Competitive analysis and autoresearch ```
GitHubで見る
この SKILL.md は非常に大きいため、SkillsMP では最初のセクションだけを表示しています。 GitHubで見る