AI PM OS vs free PM skill packs on GitHub: which should you use?

By George Nurijanian Published October 24, 2026

Which is better: AI PM OS or free GitHub skill packs?

Free PM skill packs on GitHub, such as phuryn/pm-skills, give you well-built skill files, but you choose, organise and maintain them yourself. AI PM OS ($99 one-time) is a packaged workspace: 231 skills, a 5-file context system, 11 guided workflows and 12 review agents, kept updated. Pick free to build it yourself. Pick AI PM OS to skip that work.

Key takeaways

If you mostly need PRD, discovery and strategy commands, phuryn/pm-skills covers it for free.
If you want to learn why each framework works, Dean Peters' pack is the best teacher of the lot.
None of the free packs ships a filled-in context system for your product. You write that yourself, whichever pack you use.
AI PM OS charges for assembly and upkeep: context files, workflows, reviewers and memory wired together and kept current. The skill files alone aren't the point.
You can try before paying: my free sample repo and a 30-day refund.

The free packs, fairly described

phuryn/pm-skills (PM Skills Marketplace)

Repo · MIT · ~27.3k stars, ~2.8k forks

Paweł Huryn's pack. The README lists 69 PM skills and 42 chained workflows across 9 plugins; the repo description says "100+ agentic skills, commands, and plugins." Designed for Claude Code and Cowork; Codex reads the same plugin marketplace, and the skills can be copied into Cursor or Gemini CLI. Start commands: /discover, /strategy, /write-prd. Version 2 added /red-team-prd for adversarial PRD review (write-up).

Strength: breadth, chained commands, and the largest community of any PM pack.

Dean Peters' Product-Manager-Skills

Repo · CC BY-NC-SA 4.0 · ~7.6k stars · v0.84 (10 August 2026)

77 skills for Claude Code, Cowork, Codex, Cursor, ChatGPT and other agents. Dean's stated goal is "functional and pedagogic in equal measure": each skill does the work and teaches the reasoning, with good and bad examples.

Strength: teaching. If you want to understand a framework well enough to explain it to your team, start here.

License note: the README says you can use the skills in your day job and adapt them (sharing alike), but can't repackage them into a paid product. This is a description of the license, not legal advice; read the license if you plan to build on it.

product-on-purpose/pm-skills

Repo · Apache 2.0 · ~757 stars

68 skills organised around the Triple Diamond, plus Foundation Sprint and Design Sprint tools; 10 workflow commands; sub-agents such as pm-critic; 200+ sample outputs; CI-enforced skill contracts. Installs as a Claude Code plugin, through the open skills CLI for other agents, or as an MCP server.

Strength: rigor and examples. Good for teams that want every PRD to come out in the same shape.

Anthropic Product Management plugin

Plugin page · official, free · part of knowledge-work-plugins (Apache 2.0)

Seven commands: /write-spec, /roadmap-update, /stakeholder-update, /synthesize-research, /competitive-brief, /metrics-review, /brainstorm. Built primarily for Cowork, also works in Claude Code, with MCP connectors for trackers, Slack, Notion, Figma, analytics and support tools.

Strength: official support and connectors that pull context from your tools.

gnurio/nurijanian-skills (my free sample)

Repo · MIT · ~127 stars

A small set of my own PM skills, full source, written to the same standard as AI PM OS. Install with npx nurijanian-skills; start with make-requirements-great.

What AI PM OS is

AI PM OS is a Markdown workspace you download and open in Claude Code, Codex, Cowork or Cursor. It contains 231 skills, 150+ frameworks, 11 guided workflows, 12 review agents (9 named reviewers plus 3 orchestrators), 6 MCP server configurations, a 5-file context system set up with /start, and 7 PRD templates. Memory capture (/capture-memory) records decisions and risks as you work and asks before it writes. $99 one-time through Polar, 30-day money-back guarantee, free lifetime updates. Files stay on your machine; nothing routes through prodmgmt.world. The full skill list is on the skills catalog.

Feature-by-feature comparison

Checked 24 October 2026. "Not published" means the maker doesn't say, not that the feature is missing.

Feature-by-feature comparison
Criterion	AI PM OS	phuryn/pm-skills	Dean Peters	product-on-purpose	Anthropic PM plugin
Price	$99 one-time	Free	Free	Free	Free
License	Paid license	MIT	CC BY-NC-SA 4.0	Apache 2.0	Apache 2.0 (repo)
Skills (as published)	231	69	77	68	7
Chained workflows / commands	11 workflows	42 workflows	Not published as a count	10 workflow commands	7 commands
Review agents / sub-agents	12 review agents	`/red-team-prd`, `/pre-mortem` commands	Not published	Sub-agents incl. pm-critic	Not published
Context system	5 files, guided via `/start`	You write it	You write it	You write it	Pulls from MCP connectors; otherwise you provide it
Memory capture	`/capture-memory`, confirmation-gated	Not published	Not published	Not published	Not published
Tools supported	Claude Code, Codex, Cowork, Cursor	Claude Code, Cowork; skills in Codex, Cursor, Gemini CLI	Claude Code, Codex, Cursor, ChatGPT, others	Claude Code; others via skills CLI, AGENTS.md, MCP	Cowork, Claude Code
Install	Polar download, open folder, `/start`	Plugin marketplace	Plugin marketplace or zip	Plugin marketplace, `npx skills add`, MCP	Plugin marketplace
Updates	Re-download from Polar; free for life	Active (last push 16 Oct 2026)	Active (last push 16 Oct 2026)	Active (v2.34.0, 17 Oct 2026)	Maintained by Anthropic
Support	Email the maker	GitHub issues	GitHub issues	GitHub issues	Anthropic
Best for	Context + workflows + reviewers wired	Free breadth	Learning frameworks	Consistent team artifacts	Cowork teams, official support

What you're really paying for (and what you're not)

From my own FAQ, because it's the honest version: "Free packs are a real option, and I link to good ones myself." With free packs, the assembly is on you: deciding which skills to keep, structuring folders, writing context files, wiring workflows and MCP servers, and keeping it all current as Claude Code and Codex change. "The $99 buys you out of months of assembly, not the files themselves."

The counterpoint is real too. The free packs are maintained: all three community repos had commits in the past fortnight, and Anthropic maintains its own. Assembling your own setup also teaches you how the system works, which matters if you'll be the person on your team who fixes it.

When free packs are the better choice

You need three to five specific skills, say a PRD writer and a JTBD interview guide. Install them from phuryn or pmprompt and you're done.
You enjoy tinkering and want to own every file.
Your company won't approve a paid tool without procurement, and free, open-licensed files are easier to clear.
You're learning how skills and frameworks work. Dean Peters' pack is built for that.
Your team is on Cowork and wants the vendor's own plugin. Use Anthropic's.

When a packaged OS is the better choice

You want context, workflows and reviewers working together on day one instead of after weeks of evenings.
You're a solo PM without a reviewer. The review agents are the gut-check you don't have.
You don't want to maintain it as the editors change.
You switch between editors (Claude Code at work, Cursor at home, Codex for some tasks) and want one workspace for all of them.

Can you combine them?

Yes, in general. Skills are folders, so you can add a free pack's skills alongside AI PM OS or any other setup; AI PM OS is a folder of Markdown files and doesn't replace your existing Claude or Cursor configuration. Watch for overlapping command names such as /write-prd or /brainstorm, and keep only one version of each job so you know which one ran.

Try before you decide

Install my free sample: npx nurijanian-skills, then run make-requirements-great on a real requirement.
Install one free pack (phuryn's is the quickest) and run it on the same requirement.
Score your setup with the free AI PM Setup Scorecard to see whether your gap is skills, context or review.
If the gap is context, workflows and review, AI PM OS is $99 one-time with a 30-day refund.

FAQ

Are free PM skill packs on GitHub good enough?

For many PMs, yes. phuryn/pm-skills, Dean Peters' pack and product-on-purpose/pm-skills are well made and actively maintained. What they leave to you is choosing skills, writing your context files, wiring workflows and keeping it current.

Why pay $99 for AI PM OS when free skill packs exist?

You're paying for the assembly and upkeep: 231 curated skills, a 5-file context system, 11 workflows, 12 review agents and updates maintained against current Claude, Codex and Cursor releases. If you'd rather build that yourself, the free packs are a real option.

What's the best free Claude Code skill pack for product managers?

It depends on the job. phuryn/pm-skills for breadth and ready commands, Dean Peters' Product-Manager-Skills for learning frameworks, product-on-purpose/pm-skills for consistent team artifacts, and Anthropic's Product Management plugin for official Cowork support.

Can I use phuryn/pm-skills and AI PM OS together?

Generally yes; skills are folders and can sit side by side. Avoid running two skills for the same job with the same command name, or you won't know which one answered.

What license are the free PM skill packs under, and can I use them at work?

phuryn/pm-skills is MIT and product-on-purpose/pm-skills is Apache 2.0; both allow commercial use. Dean Peters' pack is CC BY-NC-SA 4.0; its README says day-job use is fine but repackaging for sale isn't. Check each license yourself.

Do free packs include context files or review agents?

None ships a filled-in context system for your product; you write it. Some include review-style commands, such as phuryn's /red-team-prd and product-on-purpose's pm-critic sub-agent.

How do I install a PM skill pack from GitHub?

Most install through Claude Code's plugin marketplace, for example claude plugin marketplace add phuryn/pm-skills, or through npx skills add <repo> for other agents. Each README has exact commands.

Does AI PM OS work in Cursor and Codex as well as Claude Code?

Yes. It runs in Claude Code, Codex, Claude Cowork and Cursor.

Is there a refund if AI PM OS isn't worth it for me?

Yes, a 30-day money-back guarantee. Email within 30 days of purchase for a full refund.

Can I see AI PM OS skill content before buying?

Yes. The free sample at github.com/gnurio/nurijanian-skills is full skill source written to the same standard. Start with make-requirements-great.

About the author

George Nurijanian - Product Leader & AI PM OS creator

Product Leader with 8+ years of experience and the maker of AI PM OS.

Builds and maintains AI PM OS, one of the options compared on this page. Facts about the free packs come from their own READMEs, checked 24 October 2026.