Turning Product Management into an AI Operating System: A Deep Dive into pm‑skills

Overview

pm‑skills Marketplace is an open‑source AI‑driven product‑management workflow engine that encodes established product frameworks (Teresa Torres continuous discovery, Marty Cagan product strategy, Dan Olsen PRD, Alberto Savoia prototype testing) into callable AI commands.

Repository

GitHub:

https://github.com/phuryn/pm-skills

Community metrics

Stars: 8.6k+

Forks: 884

Skills: 65+

Commands: 36

Plugins: 8

Architecture

Three‑layer model:

Skill – smallest unit that implements a single product‑management task (e.g., write a PRD, generate an opportunity‑solution tree, draft a user‑interview guide).

Command – chains multiple Skills into a reusable workflow (e.g., /discover expands an idea → identifies hypotheses → ranks risks → designs experiments).

Plugin – groups related Commands for a product phase (discovery, strategy, execution, market research, data analytics, go‑to‑market, growth, toolkit).

Plugin modules

pm‑product‑discovery : 13 Skills, 5 Commands (based on Teresa Torres).

pm‑product‑strategy : 12 Skills, 5 Commands (based on Marty Cagan).

pm‑execution : 15 Skills, 10 Commands (covers PRD, OKR, roadmap, sprint planning).

pm‑market‑research : 7 Skills, 3 Commands (user personas, market segmentation, competitor analysis).

pm‑data‑analytics : 3 Skills, 3 Commands (SQL generation, queue analysis, A/B testing).

pm‑go‑to‑market : 6 Skills, 3 Commands (GTM strategy, growth loop, launch plan).

pm‑marketing‑growth : 5 Skills, 2 Commands (marketing ideas, north‑star metric).

pm‑toolkit : 4 Skills, 5 Commands (resume review, legal check, grammar check).

Typical command workflow

Example: launch a new product from idea to market.

Run /discover <product idea> – generates an opportunity tree, hypothesis list, risk ranking, and experiment design.

Validate hypotheses, then run /strategy <product domain> – produces a 9‑section strategy canvas (vision, target market, value proposition, pricing, competition, metrics, etc.).

Run /write-prd <feature description> – outputs a structured PRD with user stories, acceptance criteria, and success metrics (Dan Olsen & Cagan format).

Run /plan-launch <product name> – creates a go‑to‑market plan covering positioning, launch timeline, and growth loop.

Run /north-star <product name> – defines a hierarchical north‑star metric and supporting KPIs.

Installation

Claude Cowork (recommended)

Open Claude Cowork.

Navigate to Customize → Browse plugins → Personal → +.

Select “Add marketplace from GitHub” and enter phuryn/pm-skills.

Confirm to install all eight plugins.

Claude Code CLI

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

# Install each plugin
claude plugin install pm-toolkit@pm-skills
claude plugin install pm-product-discovery@pm-skills
claude plugin install pm-product-strategy@pm-skills
claude plugin install pm-execution@pm-skills
claude plugin install pm-market-research@pm-skills
claude plugin install pm-data-analytics@pm-skills
claude plugin install pm-go-to-market@pm-skills
claude plugin install pm-marketing-growth@pm-skills

Comparison with alternatives

Compared with a generic prompt library or a general‑purpose AI assistant, pm‑skills provides:

Core positioning : dedicated product‑management workflow engine vs. scattered prompts vs. generic assistant.

Methodology : built‑in expert frameworks vs. no fixed methodology.

Output quality : structured, executable artifacts vs. random or shallow text.

Workflow : chainable, phase‑linked processes vs. single‑shot usage.

Learning curve : requires basic product knowledge (higher than generic tools).

Platform dependency : optimized for Claude Code/Cowork (universal for others).

Open‑source degree : MIT‑licensed, fully customizable vs. partially open or closed.

Key advantages

Embedded professional methodology reduces hallucinations and keeps AI actions aligned with proven product practices.

Chainable commands let teams move from discovery to launch without manual hand‑off.

Structured outputs (PRDs, strategy canvases, GTM plans) are ready for stakeholder review.

MIT license enables modification or extension of any plugin.

Limitations

Full command set works best on Claude Code/Cowork; other platforms may lack compatibility.

Result quality depends on clear, detailed input (“garbage in, garbage out”).

Users need foundational product‑management knowledge; the system does not replace basic understanding.

Target audience

Designed for experienced product managers, technical founders or CEOs who need to craft strategy without a dedicated PM, tech leads participating in product decisions, and growth or operations professionals applying product thinking. Newcomers can use it as a guided learning aid after acquiring core concepts.

Performance impact (reported)

PRD authoring time reduced from ~4 hours to ~1 hour (including revisions).

Structured output lowers rework and improves decision quality.

Practitioners report 3–5× faster acquisition of framework knowledge when using the system as a guided practice tool.

Conclusion

pm‑skills Marketplace is a comprehensive, open‑source AI‑powered product‑management system that transforms classic frameworks into executable commands. When used with Claude Code or Cowork, it delivers end‑to‑end, structured workflows that can accelerate product development and improve decision quality, at the cost of requiring platform support and baseline product‑management expertise.