Product-Manager-Skills: a library of structured PM frameworks that equip AI agents to perform professional product management work

What it solves

It prevents AI agents from producing generic, low-quality product management (PM) deliverables. By providing a structured library of professional frameworks, it ensures that both the human PM and the AI agent share the same foundational reasoning, failure modes, and judgment, leading to more actionable and trusted outputs.

How it works

The project consists of a library of 52 "skills" delivered as structured Markdown files (SKILL.md). These files include frontmatter for agent discovery, key concepts, step-by-step application instructions, and common pitfalls. The skills are organized into three tiers:

• Workflow Skills: End-to-end processes (e.g., full discovery cycles).
• Interactive Skills: Guided discovery using an "Adaptive Decision Ladder" that asks 3-5 targeted questions before recommending a path.
• Component Skills: Templates for specific deliverables (e.g., user stories).

These skills can be integrated into various AI platforms (Claude, ChatGPT, Cursor, etc.) or run via a local Streamlit playground.

Who it’s for

Product Managers who use AI agents to assist with strategy, research, and delivery, as well as those looking to learn the reasoning behind professional PM frameworks.

Highlights

• Broad PM Coverage: Includes frameworks for problem framing, stakeholder mapping, customer discovery, prioritization, and financial health.
• AI-Specific PM Skills: Dedicated tools for context engineering, agent orchestration, and AI-readiness assessments.
• Multi-Platform Support: Ready-to-use packs for Claude Desktop/Web, Claude Code, Codex, and other LLM-based agents.
• Pedagogic Approach: Focuses on teaching the "why" behind the framework to the human user while equipping the agent to execute it professionally.