A lightweight operating system for product managers, built on Cursor IDE. It gives LLMs shared context about your products, strategy, and decisions — so every conversation starts informed.
Created by @hardiktiwari and @Sach1ng
Blog: Part 1 — Giving AI Your Context | Part 2 — Self-Updating Knowledge Base
-
Clone this repo
git clone https://github.com/hardiktiwari/PM-operating-OS.git cd PM-operating-OS -
Open in Cursor and say: "onboard"
The onboarding agent asks about your role, products, goals, and tools — then writes the config, creates your product folder, and sets up MCPs.
-
Restart Cursor to load rules.
| Folder | Purpose |
|---|---|
AGENTS.md |
Persistent memory — persona, learned preferences, workspace facts |
skills/ |
On-demand PM capabilities (PRDs, experiments, launch posts, etc.) |
.cursor/agents/ |
Onboarding agent |
knowledge/products/ |
Product dossiers — brief, roadmap, metrics, experiments, releases |
workspace/drafts/ |
Agent scratch space for drafts and analysis |
config/ |
Your configuration (written during onboarding) |
.cursor/mcp.json |
MCP connections for Figma, Google Drive, Jira |
| Workflow | Skills |
|---|---|
| Planning | strategy-connector, working-backwards |
| Building | prd-writer, one-pager, experiment-designer |
| Shipping | launch-readiness, launch-post |
| Communicating | exec-communicator, stakeholder-update |
| Learning | experiment-writeup, continual-learning |
| Operating | meeting-to-actions, brainstorming, writing-clearly |
| Decisions | decision-logger, what-if, knowledge-updater |
Each product lives in knowledge/products/<product-name>/:
knowledge/products/my-product/
├── brief.md # What it is, problem, hypothesis, scope
├── roadmap.md # Now / Next / Later + milestones
├── metrics.md # Dated metric snapshots
├── updates.md # Status updates (newest first)
├── releases/ # Per-release docs
└── experiments/ # Experiment write-ups
Copy _template/ to create a new product. The agent reads brief.md first for any product question.
Pre-configured in .cursor/mcp.json — add your credentials:
| Tool | What you need | Where to get it |
|---|---|---|
| Figma | Personal Access Token | figma.com/developers |
| Google Drive | OAuth Client ID + Secret | Google Cloud Console |
| Jira | API Token + Site URL | Atlassian API Tokens |
Or install from the Cursor marketplace: Settings → Plugins → search for Figma, Atlassian, etc.
PM-OS maintains persistent memory in AGENTS.md:
- Learned User Preferences — behavioral corrections you make (e.g., "don't assume numbers")
- Learned Workspace Facts — durable truths about your workspace (e.g., file locations, conventions)
Install the Continual Learning plugin to keep this updated automatically:
/add-plugin continual-learning
It runs after every conversation — mines transcripts for corrections and facts, writes them into AGENTS.md. No manual upkeep.
Already using PM-OS? See UPGRADING.md for how to pull new features. Check CHANGELOG.md for what changed in each release.
MIT — see LICENSE for details.
Not affiliated with any employer.