Skip to content
This repository was archived by the owner on Jul 9, 2026. It is now read-only.

Latest commit

 

History

66 Commits

Folders and files

NameName
Last commit message
Last commit date
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Oh My PM

Oh My PM is an open-source Head of Delivery agent kit for project, product, software, and marketing execution.

Oh My PM is not a prompt pack. It is a reusable delivery leadership layer for AI agents.

License: MIT Version

New here? Start with the Quickstart, pick your package from the Support matrix, or browse all documentation.


What is Oh My PM?

Oh My PM gives AI agents the thinking, judgment, and structure of a Head of Delivery.

Instead of writing the same project management instructions from scratch every time, you install Oh My PM once and your agent gains:

  • Structured delivery leadership behavior
  • Project, product, software, and marketing execution patterns
  • Token-efficient agent workflows
  • Persian/English bilingual support
  • Playbooks, templates, and structured prompts

Why it exists

AI agents are excellent at generating content but often lack delivery judgment: how to prioritize, how to assess risk, how to write stakeholder updates, how to manage scope, and how to think like a Head of Delivery.

Oh My PM fills this gap with a reusable, installable, open-source delivery layer.


Who it is for

  • Product managers using AI agents
  • Engineering leads and delivery managers
  • Teams using ChatGPT, Claude, Cursor, or Codex for delivery work
  • Anyone who needs a structured AI delivery partner

Supported tools

Tool Support
Claude Code ✅ via CLAUDE.md
Cursor ✅ via .cursor/rules/
Codex ✅ via AGENTS.md + Codex Skill
ChatGPT ✅ via ChatGPT Skill
Generic agents ✅ via AGENTS.md
MCP (alpha) ✅ Read-only server — local context, GitHub, ClickUp, Airtable, Linear, Jira, Notion connectors shipped

Quick start

Not sure which package you need? See Choose your package and the Quickstart guide. The essentials are below.

One-command CLI

The CLI package name is @oh-my-pm/cli. npm publication is not automatic; after it is published:

npx @oh-my-pm/cli@latest install claude
npx @oh-my-pm/cli@latest install cursor
npx @oh-my-pm/cli@latest install codex
npx @oh-my-pm/cli@latest install generic
npx @oh-my-pm/cli@latest skill build chatgpt
npx @oh-my-pm/cli@latest mcp doctor
npx @oh-my-pm/cli@latest doctor

From a local clone (before npm publication), build the CLI and run it directly — see docs/installation.md. The CLI delegates to the installer scripts below and never stores credentials.

Install for Claude Code

# From this repository root:
bash installers/install-claude.sh

This copies CLAUDE.md into your project root. If CLAUDE.md already exists, it will not overwrite.

To force replace with backup:

bash installers/install-claude.sh --force --backup

To preview without changing anything:

bash installers/install-claude.sh --dry-run

Install for Cursor

bash installers/install-cursor.sh

This copies .cursor/rules/*.mdc files into your project.

Install for Codex

bash installers/install-codex.sh

This copies AGENTS.md and the Codex Skill into your project.

Install as ChatGPT Skill

  1. Download the latest oh-my-pm-v1.9.1-chatgpt-skill.zip from Releases.
  2. Unzip and upload SKILL.md as a custom GPT context file, or follow the ChatGPT Skill import process.

Full per-target detail (verification, update, uninstall) is in docs/installation.md.


Documentation

Guide For
Quickstart Get running in minutes
Installation Per-target install detail
Support matrix What each target supports
MCP · MCP health Read-only live context
Security Read-only boundaries
Troubleshooting · FAQ Getting unstuck
Validation Every check command
Releases Versioning & verifying downloads
Distribution Release assets, checksums, packaging
All docs Documentation index

Bilingual Persian/English support

Oh My PM supports Persian and English as first-class workflow languages.

  • Agent output matches the user's language by default.
  • Persian is used for management, delivery, stakeholder, and decision contexts.
  • English is preserved for code, CLI, APIs, schemas, and technical identifiers.
  • Bilingual templates and prompts are provided in templates/fa/ and prompts/fa/.
  • See docs/bilingual-support.md for full details.

Release assets

Each release includes:

Asset Description
oh-my-pm-vX.Y.Z-chatgpt-skill.zip ChatGPT Skill package
oh-my-pm-vX.Y.Z-codex-skill.zip Codex Skill package
oh-my-pm-vX.Y.Z-claude-pack.zip Claude Code pack
oh-my-pm-vX.Y.Z-cursor-pack.zip Cursor pack
oh-my-pm-vX.Y.Z-generic-agent-pack.zip Generic AGENTS.md pack
checksums.txt SHA-256 checksums
validation-report.md Release validation results

MCP

MCP (Model Context Protocol) support is an optional, read-only integration layer.

  • v0.7.0 shipped a read-only MCP server alpha — local context only, no external connector.
  • v0.8.0 added a read-only GitHub Issues / Projects connector.
  • v0.9.0 added a read-only ClickUp connector.
  • v0.10.0 added a read-only Airtable connector.
  • v0.11.0 added a read-only Linear connector.
  • v0.12.0 added a read-only Jira connector.
  • v0.13.0 added a read-only Notion connector, completing the currently planned connector list. See docs/mcp-connector-roadmap.md.

Run oh-my-pm mcp doctor (add --json or --strict) to check connector health — it reports env var names only, never secret values, and makes no live calls. See docs/mcp-health.md.

MCP does not replace existing packs. Current Claude Code, Cursor, Codex, ChatGPT, and generic pack support is unaffected. See packages/mcp-server/README.md.

See docs/mcp.md for the full MCP roadmap.


Security and privacy

  • Oh My PM is read-only — no write actions in any target or MCP connector.
  • Installers make no network calls and collect no telemetry.
  • No secrets, credentials, or private data are included; diagnostics print env var names only.
  • See SECURITY.md and docs/security.md for the full security model and boundaries.

Roadmap

Version Milestone
v0.1.0-alpha Repository foundation + installable alpha packs
v0.2.0 Installer hardening + safer upgrades
v0.3.0 Bilingual FA/EN quality hardening
v0.4.0 Scenario testing + golden output evaluation
v0.5.0 Deep playbooks, templates, and examples
v0.7.0 MCP research, architecture docs, and read-only server alpha
v0.8.0–v0.13.0 PM tool connectors (GitHub, ClickUp, Airtable, Linear, Jira, Notion)
v1.0.0 Stable install contract
v1.0.1 Post-release hardening (docs, connector parity, installer safety)
v1.1.0 Skill professionalization
v1.2.0 One-command installer / CLI
v1.3.0 Skill examples & output library
v1.4.0 Read-only MCP quality I
v1.5.0 Skill evaluation system
v1.6.0 Read-only MCP quality II
v1.7.0 Skill sync & build system
v1.8.0 Documentation & adoption polish
v1.9.0 Packaging & distribution hardening
v1.9.1 Repository consistency patch

Contributing

See CONTRIBUTING.md.

Maintainers: skill, agent, and pack copies are kept consistent by the artifact sync system. Run node scripts/sync-artifacts.mjs --check to detect drift and --write to repair it. See docs/artifact-sync.md.


License

MIT — see LICENSE.

About

AI-ready project/product management toolkit for turning project chaos into structured prompts, docs, decisions, risks, and execution plans.

Topics

Resources

Code of conduct

Contributing

Security policy

Stars

1 star

Watchers

0 watching

Forks

Releases

Packages

Contributors

Languages