Skillget

PM-operating-OS

mcp
Security Audit
Pass
Health Pass
  • License — License: MIT
  • Description — Repository has a description
  • Active repo — Last push 0 days ago
  • Community trust — 90 GitHub stars
Code Pass
  • Code scan — Scanned 8 files during light audit, no dangerous patterns found
Permissions Pass
  • Permissions — No dangerous permissions requested
Purpose
This tool provides a self-serve operating system and context layer for Product Managers using AI coding assistants. It helps users set up custom workflows, agents, and skills to streamline tasks like writing PRDs and analyzing customer feedback directly within Cursor or Codex.

Security Assessment
Overall Risk: Low. The codebase does not request dangerous permissions, and a scan of 8 files found no dangerous patterns or hardcoded secrets. Because it acts as an AI workflow framework, it uses shell execution to run its setup scripts (e.g., `setup.sh`) and standard package installations. It is designed to integrate with your existing work tools, so it inherently touches proprietary documents and product data. However, it does not exhibit malicious data collection or unauthorized network requests.

Quality Assessment
The project is actively maintained, with its last update pushed today. It has earned 90 GitHub stars, indicating a solid baseline of community trust and usefulness for its target audience. It includes the standard MIT license, making it fully open source and safe for commercial and personal use. The repository is cleanly structured, well-documented, and provides clear onboarding instructions.

Verdict
Safe to use.
SUMMARY

A self-serve Cursor setup for PMs: rules, skills, and agents that adapt to your role, goals, and tools. Write PRDs, analyze VOC, plan your week, and ship faster. [Admins: Sachin Gupta, Hardik Tiwari]

README.md

PM Operating System

A PM operating system that gives your LLM shared PM context: strategy, goals, domain knowledge, workflows, and memory. It connects to the tools you already use through MCPs and supports two runtime adapters:

  • Cursor for the original rules + agents workflow
  • Codex for a native Mac app workflow built on AGENTS.md, repo skills, and .codex/config.toml

Created by @Sach1ng and @hardiktiwari

Quick start

  1. Clone the repo

    git clone https://github.com/Sach1ng/PM-operating-OS.git
cd PM-operating-OS

  2. Install setup dependencies

    pip install -r requirements.txt

  3. Choose your runtime

Cursor track

  • Open the repo in Cursor.
  • In chat, say "onboard" or "PM-OS setup".
  • The onboarding agent writes config/pm-os-config.yaml and runs setup for Cursor.
  • Restart Cursor when setup finishes.

Codex track

  • Open the repo in the Codex Mac app.
  • On first run, use the checked-in bootstrap skill by explicitly invoking $pm-os-onboarding. Plain-language "PM-OS setup" should also work, but the explicit skill call is more reliable.
  • The onboarding skill writes config/pm-os-config.yaml, then generates PM-OS-AGENTS.md, .codex/config.toml, and the rest of the repo-scoped Codex workflow skills.
  • Restart Codex if the new skills or MCP config do not appear automatically.
  • If the onboarding skill is still unavailable, use the manual setup path below, then restart Codex and reopen the repo.

Manual setup

If you do not want interactive onboarding, edit config/pm-os-config.yaml and run one of:

./scripts/setup.sh --runtime cursor
./scripts/setup.sh --runtime codex --codex-scope repo
./scripts/setup.sh --runtime both --codex-scope repo

Use --output-only to preview generated artifacts without deploying them.

What PM-OS includes

Type What it is Runtime surface
Project context Chief of staff guidance, domain framing, prioritization signals Cursor rules (.mdc) and Codex project context (PM-OS-AGENTS.md)
Skills PM capabilities such as PRDs, strategy, launch, updates, decisions Cursor skills and Codex repo/user skills
Workflow assistants Feedback analysis, planning, strategy review, exec updates, retros Cursor agents and Codex workflow skills
Knowledge layer Strategy docs, metrics, customer segments, positioning Shared across both runtimes
Memory Decision traces, feedback history, planning history, retrospectives Shared across both runtimes

Structure

Path Purpose
AGENTS.md Shared repo-level PM chief of staff guidance
.agents/skills/ Codex bootstrap skills checked into the repo; setup adds the rest here for repo-scoped Codex use
.cursor/agents/ Cursor onboarding and company research bootstrap agents
config/pm-os-config.yaml Your personalized setup answers, including runtime selection
scripts/setup.py / scripts/setup.sh Runtime-aware generator and deployer
templates/ Cursor rules/agents templates plus skill templates used during setup
skills/ Canonical PM skill source folders used by setup
knowledge/ Strategy and domain context files
memory/ PM-OS memory layer
output/ Generated preview artifacts
PM-OS-AGENTS.md Generated Codex project context file (created by setup, gitignored)
.codex/config.toml Generated project-scoped Codex MCP config (created by setup, gitignored)

Runtime model

PM-OS now has two layers:

  • Portable PM capability layer: skills/, knowledge/, memory/, workflow prompts, and config schema
  • Runtime adapter layer: Cursor deployment vs. Codex deployment

Cursor

  • Generates .mdc rules from templates/rules/
  • Generates agents from templates/agents/
  • Writes project MCP config to .cursor/mcp.json
  • Deploys to ~/.cursor/

Codex

  • Generates PM-OS-AGENTS.md from the same PM context inputs
  • Writes project MCP config to .codex/config.toml
  • Exposes PM-OS workflows as Codex skills under .agents/skills/
  • Can optionally install namespaced user-scope skills under ~/.agents/skills/

PM workflows

Workflow Primary PM-OS surface
Planning strategy-connector, working-backwards, weekly-planner
Building prd-writer, one-pager, experiment-designer
Shipping launch-readiness, launch-post
Communicating exec-communicator, stakeholder-update, exec-update-generator
Learning experiment-writeup, feedback-analyzer, retrospective
Operating meeting-to-actions, action-item-prioritizer, decision-logger, what-if, knowledge-updater

See docs/agents.md for runtime-specific workflow entrypoints.
For the Codex-specific flow, see docs/codex-workflow.md.

Knowledge layer

The knowledge/ directory holds strategy and domain context that PM-OS workflows reference:

  • knowledge/_template/ — copy and customize for your domain
  • knowledge/examples/ — worked examples using public data from Spotify, Netflix, Shopify, and Uber

Start with:

  • strategy.md
  • customer-segments.md
  • metrics-and-targets.md

Those three drive the biggest quality gains across both runtimes.

Memory layer

PM-OS stores accumulated context in memory/:

memory/
├── decisions/
├── feedback/
├── weekly-plans/
├── strategy-reviews/
├── exec-updates/
├── knowledge-snapshots/
└── learning-log/

This powers:

  • temporal awareness across runs
  • cross-workflow context reuse
  • compounding PM intelligence over time

Codex v1 limitation

continual-learning remains Cursor-first / experimental in this branch. Its current implementation depends on Cursor transcript paths and is not yet ported as a first-class Codex workflow.

Requirements

Required

  • Python 3
  • pip install -r requirements.txt
  • One of:
    • Cursor with MCP support
    • Codex Mac app

Optional

  • Slack, Google Drive, GitHub, Figma, Jira, Confluence, and other MCP-backed tools

See MCP_SETUP.md for runtime-specific MCP setup.

Manual setup notes

If you want to work without the interactive onboarding flow:

  • Edit config/pm-os-config.yaml
  • Run ./scripts/setup.sh --runtime <cursor|codex|both>
  • For Codex:
    • project context is generated into PM-OS-AGENTS.md
    • project MCP config is generated into .codex/config.toml
    • repo skills are generated or synced into .agents/skills/

Contributing

  • Add or update canonical PM skills in skills/
  • Add or update Cursor workflow templates in templates/agents/
  • Add or update Codex bootstrap skills in .agents/skills/

For local verification, see docs/development-testing.md.

Authors

GitHub Role
Sachin @Sach1ng Co-creator
Hardik @hardiktiwari Co-creator

Disclaimer

This is a personal project and is not affiliated with, endorsed by, or representative of any employer.

License

MIT — see LICENSE for details.

Reviews (0)

No results found