deanpeters/Product-Manager-Skills

What it is

Product‑Manager‑Skills is a curated library of 77 “skills” – ready‑to‑use, structured prompts and templates for AI agents (Claude, ChatGPT, Codex, etc.) that help product managers (and the AI assistants they work with) perform core product‑management tasks. Each skill contains the background, failure‑mode warnings, and a short interactive workflow (usually a few questions) that guides the agent to produce a concrete artefact such as a PRD, stakeholder map, or market‑size calculation.

Who it’s for

• Product managers who want a reliable, repeatable way to run frameworks without reinventing the wheel.
• AI‑augmented product teams that use large‑language‑model agents (Claude, OpenAI, Ollama, etc.) and need the agent to have the same domain knowledge a human PM would.
• Non‑technical PMs who prefer a plug‑and‑play download rather than writing code.

Core features

Category	Example skill	What it does	Typical time
Framing & strategy	problem‑framing‑canvas	Guides you through MITRE’s Look Inward/Outward/Reframe sequence to define the real problem.	1 hr
Stakeholder alignment	stakeholder‑mapping	Produces Power×Interest and Impact×Power grids to visualise influence and priority.	30 min
Customer discovery	discovery‑interview‑prep	Generates a Mom‑Test‑style interview plan based on research goals.	1 hr
Prioritisation	prioritization‑advisor	Asks 3‑5 context questions and recommends RICE, ICE, Kano, or another method.	15 min
Delivery	prd‑development	Outputs a structured PRD (problem, personas, solution, metrics, stories).	2‑4 days
AI‑product work	context‑engineering‑advisor	Helps you design prompt‑memory architecture for AI‑powered features.	30 min
Lifecycle & EOL	eol‑process	Walks you through a six‑phase sunset plan (decide → align → plan → …).	30‑50 min
Career coaching	director‑readiness‑advisor	Provides a checklist and interview prep for moving from PM to Director.	10‑20 min

All skills are stand‑alone – you can run any one in isolation or chain them together following the suggested weekly workflows in the README.

How to use it

1. Download a pack – the repo offers ZIP packs for Claude Desktop/Web, Claude Code, Codex, or a full‑all‑skills bundle.
2. Upload to your agent – for Claude, drop the individual skill ZIPs into Claude Skills; for Codex, place the .agents/skills folder in your project.
3. Run a skill – paste the first prompt shown in the skill’s SKILL.md (e.g., Frame this problem: …) and fill in your details.
4. Optional playground – a Streamlit app (app/) lets you try skills locally without installing them in an LLM platform.

Why it matters

• Reduces generic AI output – the same contextual knowledge is baked into the prompt, so the model produces disciplined, trustworthy artefacts.
• Encodes best‑practice frameworks – each skill documents the “why” behind a method, helping both humans and agents learn the reasoning.
• Plug‑and‑play for multiple providers – works with Anthropic, OpenAI, Ollama, and Claude, making it provider‑agnostic.

Quick start checklist

• Choose your environment (Claude Desktop, Claude Code, Codex, or the Streamlit demo).
• Download the appropriate ZIP pack.
• Follow the install guide in docs/INSTALL-*.md.
• Pick a skill that matches your current need (see the “If this is you…” table).
• Paste the starter prompt, replace placeholders, and let the agent generate the output.

---

All information is taken directly from the repository’s README; no additional features are inferred.