Configuring PM Skills Plugins for Custom Use: A Complete Technical Guide

You can configure PM Skills plugins for custom use by forking the repository, modifying the YAML front-matter in SKILL.md files, validating changes with validate_plugins.py, and reinstalling the plugin via CLI.

The phuryn/pm-skills repository hosts a modular marketplace of 9 independent product management plugins compatible with Claude Code, Codex CLI, and other AI assistants. Configuring PM Skills plugins for custom use enables you to adapt domain-specific frameworks and workflows to match your organization's internal terminology and unique project requirements.

Understanding the PM Skills Plugin Architecture

The repository follows a strict self-contained architecture where each plugin operates independently without hard references to others.

Marketplace Manifest and Plugin Structure

The central registry lives in .claude-plugin/marketplace.json, which lists all 9 plugins and their source locations. Each plugin resides in its own directory (e.g., pm-product-discovery/) and contains:

• skills/ – Markdown files containing SKILL.md front-matter with name and description fields that encode reusable PM frameworks
• commands/ – Markdown files with command front-matter (description, optional argument-hint) that chain skills into end-to-end workflows
• README.md – Human-readable documentation covering installation and usage
• .claude-plugin/plugin.json – Per-plugin manifest defining name, version, author, and keywords

The repository includes validate_plugins.py to check manifest integrity, front-matter syntax, and cross-references between commands and skills.

Skills vs. Commands

Skills are individual frameworks loaded automatically when the model detects relevant queries or when explicitly invoked via prefix (e.g., /pm-product-discovery:brainstorm-ideas-new). Commands are user-invoked workflows triggered with a leading slash (e.g., /discover, /write-prd) that orchestrate multiple skills with specific arguments.

Installing PM Skills Plugins

You can install plugins through official CLIs or manual copy methods depending on your AI assistant.

Claude Code CLI Installation

Register the marketplace once, then install specific plugins or the entire collection:

# Register the marketplace

claude plugin marketplace add phuryn/pm-skills

# Install specific plugins

claude plugin install pm-product-discovery@pm-skills
claude plugin install pm-data-analytics@pm-skills
claude plugin install pm-toolkit@pm-skills

# Install all 9 plugins at once (omit plugin name)

claude plugin install

Codex CLI Installation

The same marketplace manifest works with OpenAI's Codex CLI:

# Add the marketplace

codex plugin marketplace add phuryn/pm-skills

# Install desired plugins

codex plugin add pm-go-to-market@pm-skills
codex plugin add pm-marketing-growth@pm-skills

Manual Installation for Other AI Assistants

For Gemini, OpenCode, Cursor, or other agents, copy the skill files directly:

# Copy skills from all plugin directories to agent's skill folder

for p in pm-*/; do
  cp -r "$p/skills/"* ~/.gemini/skills/  # or ~/.opencode/skills/

done

Configuring PM Skills Plugins for Custom Use

Customization requires editing source files and validating before reinstallation.

Customizing Existing Skills

1. Clone or fork the phuryn/pm-skills repository to obtain source files
2. Edit SKILL.md front-matter in the target plugin's skills/ directory to adjust name or description fields
3. Update command front-matter in commands/ files to modify description or argument-hint values for different phrasing or additional parameters

Validating Custom Changes

Before reinstalling, run the validator to ensure compliance:

python3 validate_plugins.py

The validator reports missing required fields, name mismatches, or broken cross-references between commands and skills. Fix any errors to prevent runtime failures.

Reinstalling Modified Plugins

After validation, install from your local path:

claude plugin install ./my-custom-plugin

Example: Adding a Custom Risk Assessment Skill

Create a new skill in pm-product-discovery by adding the following file at ./pm-product-discovery/skills/risk-score/SKILL.md:

---
name: risk-score
description: |
  Computes an impact × risk score for a set of assumptions.
  Triggers: when you need to prioritize risky assumptions.
---

## Prompt

Give me a table of assumptions with their impact, risk, and computed score.

Create a command that invokes this skill at ./pm-product-discovery/commands/score-assumptions.md:

---
description: |
  Runs the **risk-score** skill and returns a sorted list.
argument-hint: |
  List of assumptions (optional)
---

**risk-score** skill

The command will:
1. Ask the skill to compute scores.
2. Sort the results descending by score.

Run python3 validate_plugins.py to confirm the new skill is recognized and the command reference is valid before reinstalling the plugin.

Summary

• The phuryn/pm-skills marketplace contains 9 self-contained plugins with independent skills and commands
• Install via claude plugin marketplace add phuryn/pm-skills or codex plugin marketplace add phuryn/pm-skills, or manually copy files for other agents
• Configure custom behavior by editing YAML front-matter in SKILL.md and command markdown files
• Always execute validate_plugins.py to check manifest integrity and cross-references before reinstalling modified plugins

Frequently Asked Questions

How do I install only specific plugins from the PM Skills marketplace?

Use the CLI install command with the plugin name and marketplace suffix: claude plugin install pm-product-discovery@pm-skills. You can install multiple specific plugins by running the command repeatedly for each desired plugin, or install all 9 at once by running claude plugin install after adding the marketplace.

What is the difference between a skill and a command in PM Skills?

A skill is a reusable framework defined in a SKILL.md file with YAML front-matter containing name and description fields, automatically loaded by the AI model. A command is a workflow defined in the commands/ directory that chains one or more skills together and is explicitly invoked by the user with a leading slash (e.g., /discover).

How do I validate my custom plugin modifications?

Run python3 validate_plugins.py from the repository root. This script checks that all .claude-plugin/plugin.json manifests are valid, verifies that SKILL.md files contain required front-matter fields (name, description), and ensures that commands correctly reference existing skills without broken cross-references.

Can I use PM Skills plugins with AI assistants other than Claude Code?

Yes. While Claude Code and Codex CLI support native plugin installation via marketplace commands, you can manually copy the skills/ directory contents from any plugin folder into your agent's skill folder (e.g., ~/.gemini/skills/ or ~/.opencode/skills/). The markdown files with front-matter are compatible with any assistant that reads skill definitions from disk.

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