Prerequisites for Using pm-skills: Essential Setup Guide for Product Managers

To use pm-skills, you need basic product management knowledge, access to a supported AI assistant like Claude Code or Codex CLI, Python 3.8+ for validation scripts, and familiarity with the plugin-skill-command architecture.

The pm-skills repository is a marketplace of nine independent plugins that expose 68 product-management skills and 42 chained commands to AI assistants. Understanding the prerequisites for using pm-skills ensures you can install, configure, and leverage these tools effectively from the first command. This guide covers the technical and domain knowledge requirements based on the repository structure and documentation found in README.md and CLAUDE.md.

Core Product Management Knowledge

Before installing the plugins, you should understand the underlying frameworks that the skills implement. The repository provides concrete implementations of standard product management methodologies.

You need familiarity with frameworks such as:

• Discovery and Lean Canvas
• SWOT analysis and Porter's Five Forces
• OKRs and North Star metrics
• Roadmapping and Go-to-Market (GTM) strategies
• Cohort analysis and other analytics methods

Understanding these concepts allows you to ask the right questions when using skills like lean-canvas or swot-analysis. Each skill in paths like pm-product-strategy/skills/lean-canvas/SKILL.md assumes you recognize the methodology it implements.

AI Assistant Access Requirements

The pm-skills marketplace is consumed as a plugin by compatible AI assistants. You must have access to one of the supported platforms to load and execute the skills.

Supported AI assistants include:

• Claude Code and Claude Cowork
• Codex CLI
• Gemini CLI
• Cursor and Kiro

These tools read the SKILL.md format and automatically load skills when they detect relevant terminology. Commands must be invoked explicitly using the forward-slash syntax, such as /discover.

Installation Commands

Add the marketplace to your AI assistant using the CLI:

# Add the marketplace (Claude Code)

claude plugin marketplace add phuryn/pm-skills

Install individual plugins as needed:

# Install specific plugin

claude plugin install pm-product-discovery@pm-skills

Force-load a specific skill when auto-detection is ambiguous:

/pm-product-discovery:lean-canvas

Run chained commands that utilize multiple skills:

/discover AI-powered meeting summarizer for remote teams

Technical Prerequisites

While the skills run within AI assistants, certain technical requirements enable full functionality and validation.

Python 3.8 or Higher

The validation scripts and test suite require Python 3.8+. These tools ensure plugin integrity and consistency with the marketplace manifest defined in .claude-plugin/marketplace.json.

Validate your installation locally:

python3 validate_plugins.py
python3 -m unittest discover -s tests

The validator checks plugin JSON, skill front-matter, command front-matter, and intra-plugin references as implemented in validate_plugins.py.

Git (Optional but Recommended)

Git enables you to clone the repository, inspect source files like pm-<plugin>/.claude-plugin/plugin.json, or contribute new skills. While optional, cloning allows you to browse raw SKILL.md files and run unit tests locally.

# Example: Copy skills to OpenCode directory

for plugin in pm-*/; do
  mkdir -p .opencode/skills/
  cp -r "$plugin/skills/"* .opencode/skills/ 2>/dev/null
done

Understanding the Plugin Architecture

Grasping the relationship between plugins, skills, and commands prevents usage errors and helps you navigate the marketplace effectively.

According to CLAUDE.md (lines 49-59), the architecture follows these rules:

• Skills are noun-based and auto-load when the assistant detects relevant terminology
• Commands are verb-based and must be invoked explicitly (e.g., /discover)
• Each plugin is installable independently without cross-plugin hard-references

This separation prevents dependency conflicts. For example, pm-product-discovery/commands/discover.md chains multiple skills but remains self-contained within its plugin.

Validation and Consistency

Running the test suite ensures your local changes remain consistent with the marketplace manifest and versioning.

The test suite in tests/ includes:

• test_validator.py - Plugin structure validation
• test_consistency.py - Marketplace manifest alignment

Execute tests to verify plugin integrity:

python -m unittest discover -s tests

Summary

• Domain knowledge in product management frameworks (Lean Canvas, OKRs, SWOT) is essential for effective skill utilization
• AI assistant access through Claude Code, Codex CLI, or compatible tools is required to consume the marketplace
• Python 3.8+ enables validation via validate_plugins.py and running the tests/ suite
• Git facilitates source inspection and contribution to the nine-plugin repository
• Architecture understanding of noun-based skills (auto-loading) versus verb-based commands (explicit invocation) ensures proper usage

Frequently Asked Questions

Do I need to be a Python developer to use pm-skills?

No. While Python 3.8+ is required for running the validator and tests in validate_plugins.py, you can use the skills without any coding knowledge. The Python requirements are only necessary if you want to validate the marketplace locally or contribute to the repository.

Can I use pm-skills with ChatGPT or other web-based AI assistants?

No. The pm-skills marketplace requires AI assistants that support the SKILL.md plugin format and CLI installation commands. Supported platforms include Claude Code, Claude Cowork, Codex CLI, Gemini CLI, Cursor, and Kiro. Standard web-based ChatGPT does not support this plugin architecture.

What happens if I don't have product management experience?

You can still install and run the commands, but the output will be less meaningful. The skills in directories like pm-product-discovery/skills/ assume familiarity with frameworks like Lean Canvas and Discovery. Without this background, you may struggle to interpret results or ask effective follow-up questions.

How do I verify my installation is working correctly?

Run the validation script and test suite:

python3 validate_plugins.py
python3 -m unittest discover -s tests

These commands check that the plugin JSON in .claude-plugin/marketplace.json aligns with the actual skill files and that all intra-plugin references are valid.

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