core-ai-pm-workflow

This is a toolkit of 11 AI-powered skills built by a PM, for PMs. Clone this repo, and your AI assistant (Claude Code or Cursor) becomes a PM productivity machine — writing docs, reviewing strategy, researching competitors, building presentations, and tracking your work across GitHub.

These aren't prompts. They're workflows. Each skill is a multi-step process that knows how to gather context, ask the right questions, and produce real artifacts you can ship.

Official Site:

https://aipmworkflow.io/index.html

What You Get

Quick Start

Prerequisites

• Claude Code or Cursor
• GitHub CLI (gh) — installed and authenticated
• Python 3.9+ (for calendar and Google Docs skills)

Install

# Clone the repo
git clone https://github.com/YOUR_USERNAME/core-ai-pm-workflow.git
cd core-ai-pm-workflow

# Open in Claude Code
claude

# Or open in Cursor
cursor .

Then say: "Start the journey" — and the AI walks you through setup and your first skill.

Or jump straight in:

"Write a one-pager for [your feature idea]"
"Deep dive into the authentication system in [org/repo]"
"Create a presentation about [your initiative]"
"Review this doc as a CTO"

The Journey

New to this? Follow the guided walkthrough. Each step builds on the last and produces a real artifact you keep.

How It Works

Each skill is a SKILL.md file — a structured prompt that tells your AI assistant exactly how to handle a task. When you say "write a one-pager," the AI reads skills/doc-writer/SKILL.md and follows a multi-step process:

1. Understand what you're asking for
2. Gather context from your codebase, docs, or data
3. Produce the artifact using proven templates and formats
4. Iterate based on your feedback

You don't need to know how the skills work. Just describe what you want in natural language.

Works with Claude Code AND Cursor

The skills system is compatible with both tools. The setup guide covers configuration for each.

Repo Structure

core-ai-pm-workflow/
├── README.md              ← You are here
├── CLAUDE.md              ← AI assistant configuration
├── skills/                ← The 11 skills
│   ├── calendar/          ← Google Calendar reader
│   ├── gdocs-share/       ← Google Docs read/write/sync
│   ├── deep-dive/         ← GitHub repo researcher
│   ├── competitor-research/ ← Competitive analysis
│   ├── doc-writer/        ← Document writer (7 formats)
│   ├── doc-reviewer/      ← Document reviewer (CTO/GM personas)
│   ├── html-reference-page/ ← HTML page generator
│   ├── presentation/      ← Slide deck builder
│   ├── morning-standup/   ← Daily metric standup
│   ├── my-initiatives/    ← GitHub issue tracker
│   └── security-review/   ← Security auditor
├── journey/               ← Guided walkthrough (8 steps)
├── templates/             ← Document format templates
├── assets/                ← Branding assets (banner, etc.)
├── docs/                  ← HTML documentation pages
├── output/                ← Generated artifacts (gitignored, created on first use)
├── tasks/                 ← Your task tracking
└── data/                  ← Persistent data and configs (gitignored, created on first use)

Skill Highlights

Doc Writer — 7 PM Document Formats

Not just "write me a doc." The doc writer understands PM document formats and infers the right one from your request:

• Concept One-Pager — New ideas, vision pieces (narrative format)
• Problem/Solution One-Pager — Feature enhancements (structured format)
• Decision One-Pager — Choosing between options (opinionated format)
• PRFAQ — Big bets (press release + FAQ format)
• PRD / Structured Epic — Full specs with work streams
• Vision Epic — Direction-setting for new initiatives
• GitHub Issue — Publish directly to any repo

Doc Reviewer — CTO and GM Personas

Get feedback from the perspective of:

• CTO — Feature fit, simplicity, green field fallacy, phasing risk
• GM (Concept/Problem-Solution) — Phase clarity (Explore/Expand/Extract), problem gravity, adoption mechanics
• GM (Decision) — Decision clarity, ownership, reversibility, opportunity cost

Based on Kent Beck's 3X framework — not generic AI feedback.

Morning Standup — Your Metrics, Your Way

Configure metrics that matter to you. The skill runs your queries, evaluates thresholds, and reports what needs attention:

🔴 Needs Attention: Conversion rate dropped to 18% (threshold: 25%)
🎉 Milestone: Feature adoption hit 6,127 accounts (target: 6,000)
🟢 All Clear: Health score 99.4%, DAU normal range

Works with any SQL-compatible data source — BigQuery, PostgreSQL, Snowflake, or anything with a CLI.

FAQ

Do I need to know how to code? No. You need to be comfortable running commands in a terminal (or letting Claude Code / Cursor do it for you). The journey walkthrough assumes zero coding experience.

Does this work with my company's data? The core skills (doc writing, reviewing, presentations, research) work out of the box. Data-dependent skills (morning standup, initiative tracking) are configurable — you bring your own queries, repos, and data sources.

Is this a course or a tool? It's a tool with a guided onboarding journey. You can skip the journey and start using skills immediately, or follow it step by step to learn the workflow.

Can I add my own skills? Yes. Create a new directory in skills/ with a SKILL.md file following the same format. The AI will pick it up automatically.

What about security?

• Google OAuth tokens are stored locally and never committed (gitignored)
• No data leaves your machine except through the APIs you explicitly configure
• All skills follow security best practices (HTML sanitization, no code execution from fetched content)

Contributing

Found a bug? Have an idea for a new skill? Open an issue or submit a PR.

License

MIT License. See LICENSE for details.

Built by a PM who got tired of doing the same things manually every week.