How to Use AI-Powered Product Management Frameworks from the pm-skills Repository

The pm-skills repository provides ready-to-use AI prompts (skills) and slash commands that cover the entire product management lifecycle, from discovery to launch, using simple markdown files that any AI platform can execute.

The pm-skills repository by phuryn is a collection of AI-powered product management frameworks designed to automate strategic tasks like value proposition design, feature prioritization, and North Star metric definition. These frameworks are stored as plain markdown files with YAML frontmatter, making them compatible with any AI chat interface or API client. By leveraging these structured prompts, product managers can generate consistent, high-quality deliverables without writing complex code.

Repository Structure: Skills vs. Commands

The repository organizes its AI-powered product management frameworks into two distinct layers: skills (the underlying AI instructions) and commands (the user-facing invocation interface).

Skill Definitions

Skills are detailed AI instructions stored in SKILL.md files throughout the repository. Each skill contains YAML frontmatter defining the task metadata, followed by step-by-step instructions that guide the AI model’s output format.

Key skill files include:

• pm-product-strategy/skills/value-proposition/SKILL.md – Designs a 6-part JTBD-based value proposition
• pm-product-discovery/skills/prioritize-features/SKILL.md – Ranks feature ideas against impact/effort criteria
• pm-marketing-growth/skills/north-star-metric/SKILL.md – Guides definition of a North Star metric

Command Wrappers

Commands are markdown files that expose skills as easy-to-type slash commands (e.g., /value-proposition). These files declare the invocation syntax, workflow steps, and argument-hint fields that help the AI request missing information.

Notable command files:

• pm-product-strategy/commands/value-proposition.md – Invokes the value proposition skill
• pm-product-discovery/commands/brainstorm.md – Starts a brainstorming session using the brainstorm-ideas-new skill

The pm-toolkit/README.md file serves as the master index, listing all available skills and commands.

How the AI Frameworks Work Under the Hood

Each AI-powered product management framework in the pm-skills repository follows a consistent four-step architecture:

1. YAML Metadata – Every skill starts with a frontmatter block containing name, description, and triggers fields. This enables AI platforms like Claude to match user intent to the appropriate skill.

2. Instruction Body – The markdown body contains concise, step-by-step prompts that specify exactly what the model should output, including templates, bullet points, and tables.

3. Argument Handling – Command files declare argument-hint fields (e.g., <product or feature>) that trigger follow-up questions if the user provides incomplete input.

4. Structured Output – Skills define the final answer format (markdown tables, specific sections, or templates) ensuring consistent deliverables across the entire toolkit.

Because everything is pure markdown, the framework is language-agnostic—any chat or API client capable of sending text prompts can execute these commands.

Practical Examples: Running AI-Powered Product Management Commands

You can invoke these frameworks using slash commands in any AI interface that supports them. Replace the placeholder text with your specific product details.

Value Proposition Generation

/value-proposition
AI-powered writing tool for non-native English speakers

Result: The AI returns a complete 6-part value proposition (Who, Why, What Before, How, What After, Alternatives) plus a one-sentence marketing statement.

Feature Prioritization

/prioritize-features
Feature ideas:
- Dark mode
- In-app chat
- Export to CSV
- Real-time analytics

Result: A prioritized list using frameworks like RICE or MoSCoW, with brief justifications for each feature recommendation.

North Star Metric Definition

/north-star-metric
Our product: collaborative whiteboard for remote teams

Result: A recommended North Star metric (e.g., "daily active boards created") together with supporting leading metrics.

Programmatic Integration

You can embed these AI-powered product management frameworks into custom scripts. The following Python example forwards slash commands to Claude’s API:

import requests
import json

def call_claude(command: str, body: str):
    payload = {
        "model": "claude-2.0",
        "prompt": f"{command}\n{body}",
        "max_tokens": 1000,
        "temperature": 0.2
    }
    response = requests.post(
        "https://api.anthropic.com/v1/complete",
        headers={"X-API-Key": "YOUR_CLAUDE_API_KEY"},
        json=payload,
    )
    return json.loads(response.text)["completion"]

# Execute the value proposition framework

print(call_claude("/value-proposition", "AI-powered note-taking app"))

The script forwards the command and arguments to the API, while the heavy lifting remains inside the markdown skill definitions stored in files like pm-product-strategy/skills/value-proposition/SKILL.md.

Summary

• The pm-skills repository provides AI-powered product management frameworks as plain markdown files with YAML frontmatter.
• Skills in SKILL.md files contain the detailed AI instructions, while commands provide the slash-command interface.
• Key frameworks include value proposition design (pm-product-strategy/skills/value-proposition/SKILL.md), feature prioritization (pm-product-discovery/skills/prioritize-features/SKILL.md), and North Star metric definition (pm-marketing-growth/skills/north-star-metric/SKILL.md).
• The architecture is language-agnostic—any AI client can execute these commands by sending the markdown content as prompts.
• You can invoke frameworks manually via slash commands or programmatically through API calls.

Frequently Asked Questions

What makes pm-skills different from standard ChatGPT prompts?

The pm-skills repository structures AI instructions as reusable, version-controlled markdown files with strict YAML metadata and output formatting rules. Unlike ad-hoc ChatGPT conversations, these frameworks guarantee consistent deliverables (like JTBD value propositions or prioritized feature lists) through standardized templates defined in files such as pm-product-strategy/skills/value-proposition/SKILL.md.

Do I need programming skills to use these AI-powered product management frameworks?

No programming is required for basic usage. Simply copy the slash commands (e.g., /value-proposition) from files like pm-product-strategy/commands/value-proposition.md and paste them into your AI chat interface. Programming skills are only needed if you want to automate the process using the API integration patterns shown in the Python examples.

Can I use these frameworks with AI models other than Claude?

Yes. The pm-skills repository is designed to be model-agnostic. The markdown-based skill definitions and command structures work with any large language model that accepts text prompts, including GPT-4, Gemini, or local models. The YAML frontmatter in each SKILL.md file helps any AI platform match user intent to the correct framework.

How do I customize a skill for my specific product management workflow?

To customize a framework, fork the repository and modify the relevant SKILL.md file (e.g., pm-product-discovery/skills/prioritize-features/SKILL.md). You can adjust the step-by-step instructions, output templates, or argument hints in the YAML frontmatter. Changes take effect immediately when you invoke the corresponding command, as the AI reads the updated markdown instructions directly.

{
  "mcpServers": {
    "instagit": {
      "command": "npx",
      "args": ["-y", "instagit@latest"]
    }
  }
}