AI Product Manager

Your agent as your AI product manager — 19 skills and 21 concepts for daily product work, from discovery to board narratives.

---

🌟 Overview

AI Product Manager is a multi-harness plugin that puts an actionable AI product management practice inside your agent: 19 workflow skills and 21 concepts covering discovery, design, architecture, governance, pricing, growth, and communication — built for daily product work, from standup decisions to board narratives. The knowledge bundle implements OKF v0.2, with provenance on every concept. Its content is a cited synthesis of private educational materials and public web knowledge, including two Kellogg Executive Education certificate programs (Professor Mohan Sawhney).

✨ What it does

Skill	Purpose
ai-pm-index	Routes requests to the right skill
ai-pm-get-context	Loads product context, references, templates, local design system
ai-pm-discover	Product discovery, JTBD, opportunity analysis
ai-pm-design	Product design, user stories, MVP/MVX definition
ai-pm-architect	ML paradigm selection, system topology, deployment patterns
ai-pm-govern	ML model governance, compliance, responsible AI
ai-pm-team	Cross-functional team topology, RACI, collaboration rhythms
ai-pm-research	User research, competitive analysis, win/loss
ai-pm-price	Pricing strategy, monetization, packaging
ai-pm-grow	PLG loops, retention, activation, expansion
ai-pm-communicate	Stakeholder influence, storytelling, alignment
ai-pm-vision	Product vision, V2MOM, strategy canvas
ai-pm-audit	Product audits, responsible AI checks, tech debt
ai-pm-evals	Eval design, golden datasets, regression gates
ai-pm-platform	API/SDK design, console, token economics
ai-pm-models	Model selection, RAG/agent reliability, prompts
ai-pm-business-case	ROI/TCO business cases, P&L impact
ai-pm-vendor	Build-vs-buy, RFPs, vendor oversight
ai-pm-transform	Multi-unit roadmaps, portfolios, board narratives

Templates (11): PRD, ADR, RACI, Postmortem, Model Card, and others

🔗 Optional Dependency: okf-abstracts

The plugin's knowledge bundle uses OKF v0.2 concepts that reference types defined in okf-abstracts (pinned to v0.1.0). This is an optional dependency — the plugin works standalone, but linking to okf-abstracts provides:

• Standardized type definitions (Skill, Concept, Template, Agent, etc.)
• OWL-style class hierarchy (foundational → core → domain → application)
• Cross-bundle concept interoperability

To enable full cross-bundle resolution, add the okf-abstracts repo to your local marketplaces or place it alongside this repo.

🚀 Install

Add the marketplace once from its git URL, then install through each harness's UI or CLI. The marketplace lives at https://www.github.com/Yoseph-Zuskin/AI-Product-Manager.

Codex

1. Add the marketplace (CLI):

   codex plugin marketplace add Yoseph-Zuskin/AI-Product-Manager

2. Install in the /plugins browser UI (marketplace tab), or via CLI:

   codex plugin add ai-product-manager@ai-product-manager

3. Restart the Codex desktop app (or reload plugins) after installing.

Claude Code

1. Open /plugin and use the Discover tab (UI), or add the marketplace:

   /plugin marketplace add Yoseph-Zuskin/AI-Product-Manager

2. Install:

   /plugin install ai-product-manager@ai-product-manager

ChatGPT (web UI)

1. Open ChatGPT → Plugins → Add plugin → Custom → paste: https://www.github.com/Yoseph-Zuskin/AI-Product-Manager
2. Trust the hooks when prompted (SessionStart, SubagentStart, UserPromptSubmit)

GitHub Copilot CLI

copilot plugin marketplace add Yoseph-Zuskin/AI-Product-Manager

Then install via /plugin. In VS Code, marketplaces can also be registered in user settings JSON:

{
  "chat.plugins.enabled": true,
  "chat.plugins.marketplaces": [
    "Yoseph-Zuskin/AI-Product-Manager"
  ]
}

Cursor

In Agent chat, run:

/add-plugin https://www.github.com/Yoseph-Zuskin/AI-Product-Manager

Then select or enable ai-product-manager when Cursor prompts.

Other Agents

Also available for: Windsurf, OpenCode, Devin, Grok, Qoder, Cline, Kiro, Pi, Gemini — see .windsurf/, .opencode/, .devin-plugin/, .grok-plugin/, .qoder-plugin/, .clinerules/, .kiro/, pi-extension/, gemini-extension.json — plus an MCP server (ai-pm-mcp/) for any MCP host.

⚡ Quick Start

After installing, start a new thread and say:

Help me create a product strategy for my AI-powered feature

The ai-pm-index skill routes to the appropriate workflow. For discovery, it invokes ai-pm-discover; for design, ai-pm-design; for architecture, ai-pm-architect, etc.

🔌 MCP Server

ai-pm-mcp/ is a stdio MCP server (19 tools covering the skill set) for any MCP host. Add it to your client config:

{
  "mcpServers": {
    "ai-pm": {
      "command": "node",
      "args": ["/path/to/AI-Product-Manager/ai-pm-mcp/mcp-server.js"]
    }
  }
}

Verify with npm install && npm test in ai-pm-mcp/. Tool contract: CONTRIBUTING.md.

📦 Knowledge Bundle

The plugin includes an OKF v0.2 knowledge bundle:

• 21 concepts — AI product strategy, ML governance, pricing, architecture patterns, team topology, discovery, vision
• 19 skills — Curated workflows with AGENTS.md routing rules (the 6 newest: evals, platform, models, business-case, vendor, transform — ship as draft pending human review)
• 11 templates — PRD, ADR, RACI, Postmortem, Model Card, and others
• Provenance — Every concept cites sources from the two Kellogg programs

All content is based on cited synthesis of private educational materials and public web knowledge (not verbatim transcripts). Sources from the two Kellogg programs are cited via [^p1-xxx] footnotes — proper scholarly attribution, not copying.

🧩 Harness Compatibility

Harness	Config
Codex	.codex-plugin/plugin.json
Claude Code	.claude-plugin/plugin.json
Cursor	.cursor/rules/ai-pm.md
Windsurf	.windsurf/rules/ai-pm.md
OpenCode	.opencode/plugins/ai-pm.mjs
OpenClaw	.openclaw/skills/ai-pm/
Devin	.devin-plugin/plugin.json
Grok	.grok-plugin/plugin.json
Qoder	.qoder-plugin/plugin.json
Pi	pi-extension/
Gemini	gemini-extension.json
Copilot	.github/plugin.json
Cline	.clinerules/ai-pm.md
Kiro	.kiro/steering/ai-pm.md

⚖️ Legal

• License: MIT — see LICENSE
• Privacy: privacy.md — local-only execution, no data collection
• Terms: terms.md — MIT licensed, no warranty

📚 Source Attribution

This plugin synthesizes concepts from two Kellogg Executive Education certificate programs by Professor Mohan Sawhney:

• AI Strategies for Business Transformations
• AI-Driven Product Strategy

Materials are for personal educational use per Emeritus terms. The plugin contains synthesized concepts, not verbatim transcripts.

✅ IP & Commercial Use Verdict

This plugin is suitable for unrestricted commercial use.

The knowledge bundle contains original synthesized frameworks — not verbatim course transcripts. Key evidence:

• Original synthesized frameworks: Decision tables (ML Paradigm Selection, Model Architecture & Deployment Topology, Infrastructure & Deployment Decisions), workflows, and templates are original syntheses with analysis
• Proper attribution via sources: All concepts cite Kellogg/Emeritus courses via [^p1-xxx] footnotes — proper scholarly attribution, not copying
• Synthesis, not transcription: Content synthesizes two courses into new frameworks (e.g., "AI Canvas 2.0's define phase deliberately keeps the solution out of view..." + "Enterprise mirror of the opportunity-analysis...") — synthesis with commentary, not transcription
• No redistributable course materials: No verbatim transcripts, slides, or copied course materials are included; named frameworks (AI Canvas 2.0, V2MOM, RWW, JTBD, AI Radar 2.0, CxDNA, CMM) appear only as original summaries with attribution

The MIT license applies to the plugin code, skills, templates, and synthesized concepts. The original Kellogg/Emeritus course materials remain under their respective terms (personal educational use only per Emeritus terms).

🤝 Contributing

See CONTRIBUTING.md for the complete contribution workflow, development setup, commit conventions, gates, harness parity, MCP, and release process.

📝 Changelog

See CHANGELOG.md for release history.