Apply 69 Product Management Skills and 42 Workflows to AI Conversations
When a new idea emerges or strategy stalls, apply validated product management frameworks step by step in AI conversations instead of simply requesting text generation. This repository is a toolkit that structures the daily challenges product managers face—from discovery to strategy, execution, launch, and growth—into organized workflows. In AI environments like Claude Code or Cowork, specific commands or naturally loaded skills provide analysis and guidance integrated into actual work flows rather than static book knowledge.
Structure of Skills, Commands, and Plugins
The repository is organized into three units: skills, commands, and plugins. Skills are the basic units containing domain knowledge or analytical frameworks for specific product management tasks, loading automatically based on conversation context. Commands are user inputs that trigger end-to-end workflows connecting multiple skills. For example, the /discover command executes four skills in sequence: idea brainstorming, hypothesis identification, prioritization, and experiment design. Plugins bundle these skills and commands by domain into installable packages. Currently, nine plugins are available for areas such as discovery, strategy, and execution, and installing the marketplace applies all of them at once.
Tools by Product Lifecycle Stage
Each plugin focuses on a specific stage of product management. The product discovery plugin includes 13 skills and 5 commands for idea brainstorming and hypothesis validation. The product strategy plugin provides 12 skills and 5 commands for vision, business model, pricing, and competitive environment analysis. The execution plugin contains 16 skills and 11 commands for daily tasks such as PRD writing, OKR setting, sprint planning, and retrospectives. Additional plugins are prepared for market research, data analysis, launch strategy, marketing and growth, and AI-based code deployment. In particular, documentation and audit skills for reviewing AI-written code help identify the gap between the intent and actual implementation of code written by AI agents.
Installation Methods and Supported Environments
Installation via Claude Cowork is recommended for non-developer users. Adding the GitHub marketplace in Cowork's customize menu and entering phuryn/pm-skills automatically installs the nine plugins. Developers can add the marketplace via the Claude Code CLI using the claude plugin marketplace add phuryn/pm-skills command and then install required plugins individually. OpenAI's Codex CLI can also read the same marketplace file for native installation, allowing workflows to be executed with natural language descriptions instead of slash commands. Other AI assistants like Gemini CLI, OpenCode, Cursor, and Kiro can use the skill files by copying them, but the command functionality is exclusive to Claude.
Pre-Adoption Considerations
This repository is based on frameworks presented by prominent figures in product management, including Teresa Torres, Marty Cagan, and Alberto Savoia. It is released under the MIT license for free use. Note that slash commands do not execute directly in Codex CLI, so workflows must be described with natural language prompts. Additionally, if the Cowork VM is unstable in a Windows environment, setting up service monitoring via the PowerShell script provided in the README can resolve the issue. Skills for reviewing AI-generated code are based on static analysis and do not catch dynamic tests or runtime errors. It is advisable to verify that your product management process and AI tool usage environment fit this structure before adopting it.