## What it does
Product Manager Prompts by deanpeters (v2.1, July 2026, MIT License) is a community-built library of 96 practical prompt assets for AI-assisted product management. The library uses three interaction modes: facilitation (AI asks 3–5 questions then builds the artifact), checkpointed co-construction (section-by-section with gates), and autonomous investigation (web research with citations, labeled Fact/Inference/Assumption). Every prompt carries an HTML comment block with pedagogic notes explaining why the prompt is shaped the way it is — visible by clicking Raw on GitHub.
Directory structure: /prompts — execution-quality one-pass artifacts (PRD, jobs-to-be-done, stakeholder map, premortem, user story, lean UX canvas, agent strategy canvas, session saver), /workshops — facilitated multi-step working sessions (battle card, PRD section-by-section, opportunity solution tree, feature investment, problem framing, painstorming), /market-intelligence — autonomous research prompts with evidence contracts, question budgets, and materiality bars (market landscape scan, competitive research snapshot, competitive intel watch, voice-of-customer miner, TAM/SAM/SOM, battle card builder, SWOT, Porter's Five Forces, Ansoff matrix), /storytelling — narrative and storyboard prompts (Research-to-Narrative Bridge, hero's journey, Starts-with-Why arcs), /loops — batch and agent-ready workflows with optional Jinja2 notation (story-splitting-loop, PRD section loop, epic-story batch, research synthesis loop, competitive watch routine, SWOT batch, market sizing loop).
The market-intelligence prompts run unattended — designed for agent workflows and scheduled runs with stop conditions. The CLAUDE.md / AGENTS.md in the repo defines the authoring contract for AI-assisted contributions.
## Install / access
Repository: https://github.com/deanpeters/product-manager-prompts
Copy-paste any prompt into ChatGPT, Claude, Gemini, or Copilot. No install required. A generated catalog at catalog/INDEX.md lists every asset with a one-line description. For agent workflows using the /loops prompts, run validate-prompts.py and generate-catalog.py after any prompt edits.
## When to use
Use /prompts when you know the situation and want execution-quality output in one pass. Use /workshops when the situation is fuzzy and you want a guided working session that ends in a finished deliverable. Use /market-intelligence when you need evidence about a market, not another meeting — these prompts do the fieldwork with citations.