69 product management skills and 42 command workflows in nine plugins, covering discovery, strategy, execution, analytics, launch and reviewing AI-built code.
A broad, tidy set of PM prompts organized into plugins; useful structure, though the skills are instructions only.
The author changed 3 of 69 skill folders and the plugin manifest or hooks after our review (noticed Oct 29, 2026). The review describes the earlier version; we recheck changed skills weekly, the most significant changes first.
PM Skills Marketplace packages product management practice as nine plugins. Discovery and strategy plugins cover brainstorming, vision, business models, pricing strategy and competitive scans. The execution plugin includes skills such as create-prd and brainstorm-okrs. A data-analytics plugin has just three skills: SQL query generation, cohort analysis and A/B test analysis. Each plugin pairs skills with chained command workflows, for example /pricing or /market-product, which run several skills in sequence. A newer pm-ai-shipping plugin is aimed at reviewing AI-built code: a correctness-focused code-review skill, an intended-vs-implemented method for finding gaps between documented intent and code, and a shipping-artifacts skill listing the docs an app should have before release. Skills average about 4.6 KB and cite the books and frameworks they draw on, such as Lean Analytics; they are instruction-only, with a few reference files and no scripts, tests or evals. Installation is by plugin (claude plugin install pm-data-analytics@pm-skills, or codex plugin add for Codex). The README flags a known issue on Windows with the Cowork VM service.
69 skills averaging 4.6 KB, mostly self-contained instructions with three reference files, no scripts, tests or evals.
Counted from the repository file tree. How we score depth.
The marketplace has three layers, and knowing which one you are using makes the output easier to predict.
Skills load automatically when your conversation matches their topic. You do not need to call them. If the agent falls back on general knowledge instead of the framework you wanted, you can force a skill with /plugin-name:skill-name or just /skill-name (Claude adds the prefix). Some skills, such as prioritization-frameworks and opportunity-solution-tree, have no command of their own. They act as references the agent draws on whenever they apply.
Commands are workflows you trigger with /command-name. Each one chains one or more skills into a single process. The README's example is /discover, which runs four skills in order: brainstorm-ideas, identify-assumptions, prioritize-assumptions, then brainstorm-experiments.
Plugins bundle related skills and commands into one installable package per PM domain.
The README's "Start Here" list gives a sensible entry point for each stage of product work:
/discover to generate ideas, surface assumptions, rank them and plan experiments./strategy./write-prd./plan-launch./north-star.When a command finishes, it suggests related commands to run next, so you can move from discovery into strategy and then into a PRD without planning the sequence yourself.
You can pass context straight after a command. The README shows /market-product B2B analytics dashboard for e-commerce managers. Plain questions also work and let the right skill load on its own, for example What retention metrics should I track for a subscription app? For pricing, /pricing designs a pricing strategy with competitive analysis and experiments, drawing on the pricing-strategy skill (pricing models, competitive analysis, willingness-to-pay and price elasticity).
The skills walk you through each framework step by step and ask for your input along the way, so plan to answer questions and review drafts rather than expecting a finished document from one prompt. The README credits Teresa Torres, Marty Cagan and Alberto Savoia as sources for the discovery, assumption-mapping and prioritization methods.
Installing the whole marketplace gives you all nine plugins. You can also install them one at a time by name:
pm-toolkitpm-product-strategy: vision, business models, pricing, macro environment analysis and competitive landscape scanning (12 skills, 5 commands).pm-product-discovery: the home of the /discover chain and its idea, assumption and experiment skills.pm-market-researchpm-data-analytics: SQL query generation, cohort analysis and A/B test analysis (3 skills, 3 commands).pm-marketing-growthpm-go-to-marketpm-execution: day-to-day delivery work.pm-ai-shipping: review of AI-built code before release.The execution plugin has the most everyday PM documents. Skills in its folder include:
create-prd for product requirements documents, one of the longer skills at about 7.3 KB.brainstorm-okrs for drafting objectives and key results.outcome-roadmap for roadmaps framed around outcomes.job-stories for writing job stories.prioritization-frameworks as a standalone reference for ranking work.pre-mortem for listing failure modes before a project starts.retro for running retrospectives.release-notes for drafting release notes.dummy-dataset for generating sample data.The analytics plugin is small. If your work centers on SQL, cohorts or experiment readouts, it covers those three tasks. For broader analytics you will be combining it with your own prompts. Skills in the analytics plugin cite their sources, including Alistair Croll and Benjamin Yoskovitz's Lean Analytics.
The plugins match product management rather than project management. Searches for "pm skills" also return guides to project management skills such as scheduling and budgeting. This repo does not cover that ground. It is about deciding what to build, writing it down and checking it ships correctly.
The pm-ai-shipping plugin is aimed at founders and PMs who ship apps written by coding agents. Its three skills are meant to be used together, in this order.
1. shipping-artifacts: write intent down first. This skill defines a small core document set every app needs: architecture, user and permission flows, permissions, variables and secrets, and a test-coverage map. It adds conditional documents only when they apply: emails, scheduled work, SEO, and embedded agents or automation. Before writing anything, it looks for your repo's existing setup, including AGENTS.md, CLAUDE.md, an existing docs/ or documentation/ folder, documents that already cover the same ground under another name (such as ARCHITECTURE.md or docs/env.md), an existing coverage map such as docs/test-coverage.md, and wherever you keep audit reports. Your repo's conventions take priority over the plugin's defaults.
2. intended-vs-implemented: compare docs to code. This method treats files like permissions.md, architecture.md and variables.md as claims to verify. It reads the code that should enforce each claim and counts only a cited file and line as evidence. It checks each documented rule one boundary at a time, asking whether a server-side enforcement point covers every path. It distrusts comments such as "admin only" or "validated elsewhere" until the code confirms them. The skill's examples of what it targets include a permission that is documented but never enforced, a "cron-only" endpoint anyone can call, and a public field that leaks private data. If the docs are missing or stale, the skill reports that as the first finding and recommends documenting before auditing.
3. code-review: find defects across boundaries. Correctness is the default dimension. Performance and security are optional sub-cases that use the same process with a different anchor. The skill reviews agreements between two parts of a system, such as a caller and a callee or a writer and a later reader, rather than reviewing one file at a time. Each finding must include a feasible trigger, a concrete contradiction, an observable consequence and the strongest counterargument already checked. One root cause with both correctness and security impact gets reported once. Invocation examples from the skill:
/pm-ai-shipping:code-review
/pm-ai-shipping:code-review dimensions=correctness scope=changes
The plugin also includes a derive-tests command that groups its output into existing coverage, proposed tests and gaps where a documented rule has no test. The README's example screenshot shows a /ship-check command as well.
These are instructions, not scanners. The agent does the reading and reasoning, so results depend on how much of the codebase it can see and how complete your documents are.
Pick the right install path. The README recommends Claude Cowork for non-developers: open Customize (bottom left), go to Browse plugins, then Personal, then +, choose Add marketplace from GitHub and enter phuryn/pm-skills. All nine plugins install automatically, with commands and skills. In Claude Code, first run claude plugin marketplace add phuryn/pm-skills, then install only the plugins you want. Codex reads the same marketplace file, so codex plugin marketplace add phuryn/pm-skills followed by codex plugin add <plugin>@pm-skills works without converting or copying files.
Install selectively in the CLI. If you only write PRDs and OKRs, pm-execution alone keeps the skill list short and lowers the chance of an unrelated skill loading into a conversation. Add pm-ai-shipping separately if you also review agent-written code.
Other assistants. The README says the marketplace is designed for Claude Code and Cowork and that the skills work with other AI assistants. The plugin install commands it documents are for Claude Code and Codex.
Check the repo name. Another repository, tarunccet/pm-skills, uses the same name and also describes itself as 100+ agentic skills for product managers. The install commands on this page reference phuryn/pm-skills, curated by Paweł Huryn of The Product Compass Newsletter. Make sure you add the marketplace you mean.
Companion repos. The README links three companions from the same author: pm-brain, burnstop and claude-usage. They are separate projects, and you do not need them to use the skills.
We scanned all 161 files in this repository and read every script. Nothing in it reaches beyond what the skill describes.
Product-management skills, commands and plugin manifests that are almost entirely markdown, with a maintainer-only validator script. The static-audit commands are limited to read-only git commands and writing under reports/. The README includes an optional Windows workaround that creates a SYSTEM-level scheduled task, which you would run yourself.
The repository (161 files) on Oct 29, 2026. We had every file and read every script.
NVIDIA SkillSpector static scan: 5 pattern hits (Excessive Agency 4, analysis-evasion 1). The three scanner hits are phrases in command prose ('no verification', 'without checking') and a MEDIUM analysis-evasion flag on a plain prompt file with nothing hidden. The README also contains a PowerShell snippet for users to register a SYSTEM scheduled task as a Windows/Cowork workaround; it is troubleshooting text the user runs themselves, not something the skills execute.
A review, not a certification. Skills change after we check them — read the files before installing.
MIT — commercial use allowed.
Not really. These are conversational planning and analysis skills that need your input and review.
77 product-management skills (strategy, discovery, stakeholder work, roadmaps, PRDs) that interview you and apply a named framework, plus a few command workflows.
A large set of short design-practice skills and slash commands covering research, systems, UX strategy, UI, interaction, prototyping and design ops.
About 65 skills that apply frameworks from well-known business, marketing, UX and software books, plus guided multi-step journeys that keep state in your docs folder.
120 marketing skills across seven disciplines (narrative, SEO/GEO, social, email, paid ads, influencer, launch) with scored auditor gates and an optional 8-bot team mode.
It is a collection of 69 product management skills and 42 command workflows, grouped into nine plugins for discovery, strategy, execution, analytics, launch and AI-shipping review.
Add the marketplace to Claude Code, then install a plugin such as claude plugin install pm-data-analytics@pm-skills. For Codex, use codex plugin add pm-data-analytics@pm-skills.
Not according to the README excerpts we read. The skills produce documents, analyses and queries inside your agent session.
Both give agents PM frameworks. phuryn/pm-skills is organized as plugins with chained command workflows, while deanpeters/Product-Manager-Skills is a separate framework collection worth comparing for the specific methods you use.
Skills load automatically when the conversation matches them. Commands such as /discover or /write-prd are optional workflows that chain several skills. If the agent ignores a skill, force it with /plugin-name:skill-name or /skill-name.
Yes, in Claude Code and Codex. Add the marketplace, then install plugins individually, for example pm-execution@pm-skills or pm-ai-shipping@pm-skills. The Cowork route installs all nine at once.
No. They are separate GitHub repositories with the same name. This page covers phuryn/pm-skills, installed with claude plugin marketplace add phuryn/pm-skills.