The pm-skills repository is a comprehensive marketplace of agentic skills, commands, and plugins designed to enhance product management workflows through structured AI-assisted decision-making. The project provides over 100 skills and chained workflows organized across 9 plugins, covering the full product management lifecycle from discovery through strategy, execution, launch, and growth. It is specifically designed for Claude Code and Claude Cowork but maintains compatibility with other AI assistants through a universal skill format.
The repository addresses a core problem in AI-assisted product management: generic AI outputs lack the structure and rigor that proven PM frameworks provide. Rather than receiving unstructured text, users get step-by-step guided workflows that encode established methodologies from recognized PM thought leaders including Teresa Torres, Marty Cagan, and Alberto Savoia. This approach embeds discovery frameworks, assumption mapping, prioritization matrices, and strategic analysis directly into daily workflows rather than requiring practitioners to consult external resources.
The architecture organizes functionality into three interconnected layers. Skills serve as the foundational building blocks, each encoding domain knowledge or analytical frameworks for specific PM tasks. Commands are user-triggered workflows invoked with slash notation like /discover or /strategy, which chain multiple skills into end-to-end processes. Plugins group related skills and commands into installable packages covering distinct PM domains. This layered design allows skills to be reused across multiple commands while maintaining clear separation of concerns.
The discovery plugin contains 13 skills and 5 commands covering ideation, assumption testing, and customer research. It includes tools for multi-perspective brainstorming, assumption identification and prioritization using impact-risk matrices, opportunity solution tree construction following Teresa Torres' methodology, customer interview script generation with JTBD probing, and metrics dashboard design. The strategy plugin provides 12 skills and 5 commands addressing vision crafting, business models, pricing analysis, and competitive landscape assessment through frameworks including the Product Strategy Canvas, Lean Canvas, Business Model Canvas, SWOT analysis, PESTLE analysis, and Porter's Five Forces.
Installation varies by platform. Claude Cowork users can install directly through the plugin marketplace by entering the GitHub repository path. Claude Code users access the marketplace through CLI installation. The repository also supports other AI assistants including Gemini CLI, OpenCode, Cursor, and Kiro through universal skill file formats, though slash commands remain Claude-specific. The project includes companion repositories for related functionality: pm-brain, burnstop, and claude-usage.
GitGenius tracking data reveals the repository has grown to 22,479 stars as of the most recent check, with a median issue and PR response latency of 0.2 hours across tracked items. The most active contributors tracked include ConversationCommander, abdulrbasit, and dempseydata. The repository maintains connections to related projects including koala73/worldmonitor, dograh-hq/dograh, and panniantong/agent-reach through overlapping contributor networks.
The README emphasizes that the marketplace delivers structure over generic output, with each skill encoding proven PM frameworks and walking users through them systematically. Commands are designed to flow into one another, matching natural PM workflows and suggesting relevant next steps upon completion. The project welcomes contributions and maintains MIT licensing, with test automation and clear contribution guidelines documented in the repository.
