产品经理的 AI 技能市场：100+ 结构化工作流，从发现到增长

PM Skills Marketplace is not a chatbot wrapper—it is an AI operating system tailored for product managers. It packages over 100 proven product management frameworks (from discovery, strategy, execution to growth) into structured skills and commands that AI assistants like Claude Code can execute step by step. Instead of prompting an LLM from scratch, you get rigorous, repeatable workflows inspired by thinkers like Teresa Torres, Marty Cagan, and Alberto Savoia.

Generic AI assistants give you text; PM Skills Marketplace gives you structure. The core problem it solves is translating classic product methodologies—often stuck on bookshelves—into interactive, AI-driven daily workflows. Whether you need to map assumptions, build an Opportunity Solution Tree, or write a standardized PRD, each skill encodes a proven framework. The goal is not just faster documentation, but fundamentally better product decisions.

The marketplace is built on three layers. At the base are Skills, which encode domain knowledge or analytical frameworks and load automatically when relevant. Next are Commands (triggered by /), which chain multiple skills into end-to-end workflows. For example, /discover chains four skills: brainstorming, assumption mapping, prioritization, and experiment design. At the top, Plugins bundle related skills and commands into installable packages covering domains like discovery, strategy, and execution.

Nine plugins provide 100+ skills and 40+ commands across the product lifecycle:

• Discovery: Brainstorming, assumption testing, Opportunity Solution Trees, interviews.
• Strategy: Vision canvases, business models, pricing, competitive analysis.
• Execution: PRDs, OKRs, retrospectives, backlog breakdowns.
• Market & User Research: Personas, journey maps, TAM/SAM/SOM.
• Analytics: SQL generation, cohort analysis, A/B test evaluation.
• Go-to-Market: GTM strategy, growth loops, competitive battlecards.
• Marketing & Growth: Positioning, North Star metric definition.
• AI Artifact Review: Documenting intent for AI-built code and auditing it for security and performance gaps.

Installation depends on your tool:

Claude Cowork (recommended for non-developers)

1. Open Customize (bottom-left).
2. Go to Browse plugins → Personal → +.
3. Select Add marketplace from GitHub.
4. Enter: phuryn/pm-skills.

All 9 plugins install automatically.

Claude Code (CLI)

# Step 1: Add the marketplace
claude plugin marketplace add phuryn/pm-skills

# Step 2: Install individual plugins (example)
claude plugin install pm-toolkit@pm-skills
claude plugin install pm-product-strategy@pm-skills
# ... install the rest

Codex CLI (OpenAI)

codex plugin marketplace add phuryn/pm-skills
codex plugin add pm-toolkit@pm-skills
codex plugin add pm-product-strategy@pm-skills
# ... install the rest

Note: / slash commands won't execute in Codex; describe workflows in plain language instead.

Once installed, you can drive your work with commands:

• Got a new idea? → /discover
• Need strategic clarity? → /strategy
• Writing a PRD? → /write-prd
• Planning a launch? → /plan-launch
• Defining key metrics? → /north-star

These chain multiple skills together. For example, running /discover AI meeting assistant guides you through a full discovery cycle from brainstorming to experiment design. After each command, it also intelligently suggests the next relevant command.

Ideal For: PMs who want AI assistance without crafting fragmented prompts every time. It's perfect for individuals or small teams looking to quickly generate structured thinking (SWOTs, PRD drafts) and embed methodologies from thought leaders like Torres or Cagan directly into their daily workflow.

Caveats & Boundaries:

1. Claude Ecosystem Dependency: / slash commands are Claude-specific. On tools like Cursor or Gemini CLI, only the underlying skill files work; commands do not execute. Codex can install all plugins but cannot run the slash commands directly.
2. Not 'Magic': The skills provide rigorous, guided workflows, but the output quality still heavily depends on the quality of the inputs and context you provide.
3. Known Windows Issue: Cowork on Windows may experience VM service instability, requiring a scheduled task or manual service restart to work reliably.