DEV Community

Cover image for Open Source Project of the Day (#91): PM Skills Marketplace - Encoding Top PM Frameworks Into Your AI Agent
WonderLab
WonderLab

Posted on

Open Source Project of the Day (#91): PM Skills Marketplace - Encoding Top PM Frameworks Into Your AI Agent

#ai #opensource #agents #claude

Introduction

"Generic AI gives you text. PM Skills gives you structure."

This is article #91 in the Open Source Project of the Day series. Today's project is PM Skills Marketplace — an open-source collection that encodes top product management frameworks into AI agent skills.

You've probably tried asking AI to write a PRD, run a competitive analysis, or draft a product strategy. The output is usually fluent, well-formatted, and completely hollow — readable sentences with no real framework behind them. It doesn't look like the work of someone who's shipped products. It looks like a confident summary of nothing.

PM Skills Marketplace was built by Paweł Huryn (editor of The Product Compass newsletter) with a specific thesis: encode the actual frameworks into skill files, so the AI is forced to use the right structure when answering product questions. Not prompt engineering. Teresa Torres's Opportunity Solution Tree, Marty Cagan's INSPIRED approach, Alberto Savoia's Pretotype — all of it packaged as callable AI skills.

What You'll Learn

  • The three-layer architecture: how Skills, Commands, and Plugins work together
  • All 9 domain plugins: Discovery, Strategy, Execution, Market Research, and more
  • How to use key commands: /discover, /write-prd, /strategy, etc.
  • Which PM methodologies are encoded and from which books
  • Installation in Claude Code, Cursor, Codex CLI, and other AI tools
  • The companion PM Brain "second brain" project

Prerequisites

  • Product managers, developer-founders, or anyone who makes product decisions
  • Experience with Claude Code, Cursor, or a similar AI tool
  • Familiarity with product discovery, PRDs, or product strategy (basic level fine)

Project Background

What Is PM Skills Marketplace?

PM Skills Marketplace is a structured AI skill system designed for product managers, positioned as "The AI Operating System for Better Product Decisions."

The core problem it addresses: structured output vs. generic text. When you ask AI to do an opportunity analysis, standard AI gives you paragraphs. With PM Skills, the AI automatically applies Teresa Torres's Opportunity Solution Tree — outputting in the layered structure of Desired Outcomes → Opportunities → Solutions.

The underlying logic: the quality of product decisions depends heavily on which framework you use, and frameworks can be encoded.

Author / Team

  • Author: Paweł Huryn
  • Background: Editor of The Product Compass newsletter, senior product manager
  • License: MIT
  • Version: v2.0.0 (June 2026)

Project Stats

  • ⭐ GitHub Stars: 13,500+
  • 🍴 Forks: 1,500+
  • 📦 Content: 68 skills + 42 commands + 9 plugins
  • 📄 License: MIT

Core Features

What It Does

PM Skills creates a clear value chain:

Generic AI prompt  →  Generic text output
                           ↓ (no framework, no structure)
                       ≈ readable but useless

PM Skills prompt   →  Framework auto-activates  →  Structured product output
                                                          ↓
                                              Deliverable ready for decisions
Enter fullscreen mode Exit fullscreen mode

It moves decades of accumulated PM methodology from books and conference talks into an AI workflow.

Use Cases

  1. Product discovery

    • /discover applies the Opportunity Solution Tree to explore product directions
    • /interview generates a structured user interview guide — not just a list of questions

  2. Product strategy

    • /strategy outputs a complete product strategy document with Porter's Five Forces and Ansoff Matrix analysis
    • /market-scan runs a competitive landscape scan

  3. PRD writing

    • /write-prd generates a standards-compliant PRD with a built-in pre-mortem and red-team review

  4. Data analysis and A/B testing

    • /write-query generates BigQuery / PostgreSQL / MySQL queries
    • /analyze-test interprets A/B test results with statistical framing

  5. Go-to-market planning

    • /plan-launch produces a GTM plan including beachhead segment and growth loops
    • /north-star derives the north star metric from your business model

  6. AI code review

    • /ship-check audits AI-generated code for intent drift — catching the gap between what was intended and what was actually built

Quick Start

Install in Claude Code:

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

# Install the plugins you need (good starting set)
claude plugin install pm-product-discovery@pm-skills
claude plugin install pm-execution@pm-skills
claude plugin install pm-product-strategy@pm-skills
Enter fullscreen mode Exit fullscreen mode

Then use slash commands directly in conversation:

/discover I'm considering an AI code review tool for indie developers

/write-prd Users can trigger a one-click AI code review from VS Code,
           returning suggestions across security, performance, and readability

/strategy Create a product strategy for this tool,
          targeting indie developers who write more than 100 hours of code per month
Enter fullscreen mode Exit fullscreen mode

Install via Claude Cowork (non-developers):

Customize → Browse plugins → Add marketplace from GitHub
→ Enter: phuryn/pm-skills
→ Select plugins to install
Enter fullscreen mode Exit fullscreen mode

Other AI tools:

Tool Install path Command support
Claude Code CLI install ✅ Full
Codex CLI CLI install ⚠️ Skills only (no slash commands)
Gemini CLI ~/.gemini/skills/ ⚠️ Skills only
Cursor .cursor/skills/ ⚠️ Skills only
OpenCode .opencode/skills/ ⚠️ Skills only

All 9 Plugins

Plugin Skills Commands Core purpose
pm-product-discovery 13 5 Opportunity discovery, assumption testing, user interviews
pm-product-strategy 12 5 Product strategy, market positioning, pricing
pm-execution 16 11 PRD writing, sprint planning, risk assessment
pm-market-research 7 3 Market sizing, competitive analysis, customer journey
pm-data-analytics 3 3 SQL queries, A/B test analysis
pm-go-to-market 6 3 Launch planning, sales battlecards
pm-marketing-growth 5 2 Growth strategy, north star metric
pm-toolkit 4 5 Resume review, proofreading, legal docs
pm-ai-shipping 2 5 AI code intent audit, static security review

Deep Dive

Three-Layer Architecture: Skills / Commands / Plugins 

Plugins
  └─ Domain-specific bundles of skills — install one to cover an entire PM discipline

Commands
  └─ Triggered with /command-name — chains multiple Skills into a complete workflow
     Example: /write-prd invokes create-prd → pre-mortem → strategy-red-team in sequence

Skills
  └─ The base layer — auto-loaded into context
     Encodes specific PM frameworks with definitions and output templates
     Example: opportunity-solution-tree (includes full OST structure + output format)
Enter fullscreen mode Exit fullscreen mode

Skills load automatically — once installed, the AI uses the framework knowledge in relevant conversations without being explicitly asked. Commands are triggered manually — for complete workflows that need structured, complete deliverables.

The Encoded Methodologies

PM Skills doesn't invent frameworks — it systematically encodes the ones the field has already validated:

Product Discovery:

  • Opportunity Solution Tree (Teresa Torres) — Desired Outcomes → Opportunities → Solutions tree structure; forces you to separate problem space from solution space
  • Assumption Prioritization (Alberto Savoia) — identify and test the riskiest assumptions before building

Product Strategy:

  • Porter's Five Forces — industry competitive dynamics (existing rivals, new entrants, substitutes, suppliers, buyers)
  • Ansoff Matrix — Market Penetration / Market Development / Product Development / Diversification
  • Lean Canvas (Ash Maurya) — rapid business model assumption mapping, one page

Product Execution:

  • Pre-mortem — "assume the product failed; what went wrong?" — run before launch, not after
  • Strategy Red Team — attack your own strategy from an adversarial position to find blind spots

Metrics and Growth:

  • North Star Metric (Sean Ellis) — the single metric that best captures the core value your product delivers
  • Growth Loops — self-reinforcing growth engines, as opposed to linear acquisition funnels

OKRs:

  • Radical Focus (Christina Wodtke) — structured approach to setting and tracking Objectives and Key Results

pm-execution: The Most-Used Plugin in Detail

Execution has 16 skills and 11 commands — the domain where PMs spend most of their time.

The /write-prd workflow:

Input: /write-prd [feature description]
              ↓
Step 1: create-prd         → Generates PRD body (context / goals / user stories / acceptance criteria)
              ↓
Step 2: pre-mortem         → Assumes the feature shipped and failed; lists every plausible failure mode
              ↓
Step 3: strategy-red-team  → Critiques the PRD's strategic assumptions from an adversarial position
              ↓
Output: Complete PRD with built-in self-critique and risk assessment
Enter fullscreen mode Exit fullscreen mode

This workflow mirrors how mature product teams actually operate — a good PRD doesn't just describe what to build, it argues why this is the right thing to build and anticipates the ways it might be wrong.

The /sprint command: Breaks a feature list into sprint tasks, automatically considering dependencies and priority, and outputs a task list formatted for direct import into Jira or Linear.

pm-ai-shipping: The New Plugin Built for the AI Era

Added in v2.0.0, this plugin addresses a problem that didn't exist before AI-generated code became common:

"AI Agents write code fast but leave no record of intent."

When a feature is entirely AI-generated, teams face a new failure mode: drift between the designer's intended behavior and the AI's actual implementation. Traditional code review can't catch this because the code is syntactically correct — it just doesn't do what you meant.

The /ship-check command:

Input: feature description + AI-generated code
              ↓
intended-vs-implemented → Compares "what was meant" vs "what was built"
              ↓
shipping-artifacts      → Produces auditable artifacts (intent record + implementation notes)
              ↓
Output: Drift report + list of items requiring human sign-off
Enter fullscreen mode Exit fullscreen mode

The /security-audit-static command: Static security audit focused on AI-specific failure patterns — missing input validation, hardcoded credentials, insecure random number generation, and other mistakes AI tools make with statistically higher frequency than humans.

PM Brain: The Local Knowledge Base Companion

The companion project phuryn/pm-brain is a "product second brain" made of local Markdown files:

~/.pm-brain/
  ├── decisions/      ← Historical product decision records
  ├── learnings/      ← User interview findings and experiment conclusions
  ├── frameworks/     ← Custom methodology notes
  └── competitors/    ← Competitive tracking files
Enter fullscreen mode Exit fullscreen mode

When used together with PM Skills, the AI can reference your past decisions and product context when generating recommendations — rather than starting cold every session. No vector database. No cloud service. Plain Markdown files.

Links and Resources

Official Resources

Source Methodologies

  • Continuous Discovery Habits — Teresa Torres
  • INSPIRED & TRANSFORMED — Marty Cagan
  • The Right It — Alberto Savoia
  • Running Lean — Ash Maurya
  • Radical Focus — Christina Wodtke

Conclusion

PM Skills Marketplace does one thing: repackages decades of accumulated product management best practice in a form AI can execute.

For product managers, it doesn't answer "can AI help me do product work?" — it answers "is the AI using the right framework when it helps?" That distinction matters enormously. The first is a capability question; the second is a structure question. And structure is encodable.

For indie developers and founders, PM Skills is also a useful product thinking supplement — install a few plugins, and you can walk through the full product lifecycle from discovery to launch with AI support, without hiring a PM.

Explore PrimeSkills — A marketplace for handpicked AI Agents and skills. Each is validated in real enterprise workflows, stripping away hype and keeping only what truly works.

Welcome to my Homepage for more useful insights and interesting products.

Top comments (0)