A free operating system for Product Managers using Claude Code. One setup. Your product, your users, your standards — baked in. Every output already calibrated to you.
Most PMs use Claude like a search engine — paste a question, get a generic answer, spend an hour fixing it. PM OS changes the starting point.
Not a chatbot wrapper. A structured system built on how PMs actually work.
No coding required. Read the official Claude Code quickstart →
Open your terminal and run:
curl -fsSL https://claude.ai/install.sh | bash
Then authenticate:
claude
git clone https://github.com/derderiansevan/sevans-product-os.git cd sevans-product-os cursor .
Open CLAUDE.md and fill in every [FILL IN] field. The more specific you are, the better every skill performs.
# PM Context Role: Senior PM at Acme Corp Product: B2B SaaS for procurement teams Target users: Legal Ops Managers, mid-market OKRs: NRR 115% / churn <8% Terminology: "workspace" not "project"
In Cursor's chat panel, type a slash command:
Claude asks 2–3 clarifying questions, then generates a full PRD — hypothesis-driven, stage-appropriate, with a built-in checklist.
After generating any document:
Review this as an Engineer and a Skeptic.
Claude adopts each persona and flags specific blockers, risks, and gaps.
Each skill is a slash command. Trigger it, answer 2–3 questions, get structured output calibrated to your context.
From one-line speclet to full solution review. Adapts to your stage — Team Kickoff, Planning Review, Launch Readiness. Includes hypothesis, problem evidence, success metrics with baselines, non-goals, and rollout plan.
INVEST-compliant user stories with Given/When/Then acceptance criteria. Handles a single feature or decomposes a full product brief into a batch. Optionally posts directly to any GitHub repo.
Structured briefs for your marketing team: pain statement with customer voice, solution explanation, personas, capabilities, competitive positioning, objection handling, and key numbers.
Discussion guides for user research sessions. Warm-up → core topic exploration → hypothesis challenge → wrap-up. Includes open-ended questions designed to surface the unexpected, not confirm what you already believe.
Structured transcript summary: TL;DR, Said vs. Did table, key moments with exact quotes, surprises, and product implications. Captures signals you'd miss when skimming.
Synthesizes multiple interview transcripts into structured findings ranked by evidence strength. Distinguishes what participants said vs. did. Includes confidence levels and prioritized product implications.
Evidence-based competitor profiles: what they built, three smart decisions, three gaps, and a table of implications (copy / avoid / differentiate). Uses real data — not opinions.
Customer-facing Help Center articles. Functional, direct, no marketing language. Covers feature explanations, step-by-step guides, and limitation callouts. Adapts length to topic complexity.
Structured Slack updates for leadership and stakeholders. Format: status, wins, risks/blockers, next week's focus, asks. Skimmable, outcome-focused, and calibrated to your audience.
RICE and ICE scoring for your initiative list. Outputs Now/Next/Later format with a narrative for each horizon. Three versions: internal team, executive, and customer-facing. Requires a Not Doing list.
Three formats from one input: customer-facing (benefit-led), internal/team (technical detail), and public changelog. Works from a PRD, feature list, or GitHub diff. No feature-first language.
Slide narrative architecture for QBRs, board updates, roadmap reviews, and launch readiness. Outputs a slide-by-slide structure with talking points — so you stay in control of the story.
Slide decks for team audiences — sprint reviews, planning kickoffs, design reviews, retrospectives, and cross-functional syncs. Built for discussion, not just approval. Includes speaker notes and timed discussion prompts.
Say "review as [role]" after generating any document. Claude fully adopts the persona — lens, red flags, and output format.
Curated skills and videos to level up your PM workflow beyond this OS.
28 AI-powered skills built from 300+ episodes of Lenny's Podcast — featuring Brian Chesky, Shreyas Doshi, Marty Cagan, Kevin Weil, and more.
The exact Claude Code configuration that makes the difference — settings, context, and workflows most tutorials skip entirely.
Watch on YouTube →