How to Run Tests for phuryn/pm-skills: Integration Testing Guide

The phuryn/pm-skills repository does not include a traditional automated test suite such as pytest or unittest; instead, you validate the marketplace by installing the plugins in an AI assistant environment and executing slash commands like /discover to verify end-to-end functionality.

The phuryn/pm-skills repository is a collection of AI assistant plugins designed for product management workflows. Unlike conventional Python projects, it contains no test_*.py modules, setup.cfg configurations, or CI workflows for unit testing. Testing this marketplace requires a manual integration approach where you install the plugins and invoke skills directly within supported AI environments.

Why There Is No Traditional Test Suite

A thorough search of the codebase reveals no automated testing infrastructure. There are no test_*.py files, no pyproject.toml test configurations, and no GitHub Actions workflows executing unit tests. This is intentional because the repository consists of markdown-based skill definitions and command specifications rather than executable Python logic that requires traditional unit testing.

The skills are defined in files such as pm-product-discovery/skills/brainstorm-ideas-new/SKILL.md and pm-product-discovery/skills/prioritize-assumptions/SKILL.md, which contain prompts and instructions for AI models rather than code algorithms.

How to Run Tests for phuryn/pm-skills

Since the repository lacks conventional tests, "testing" means performing integration validation by running the actual commands the plugins provide.

Install the Marketplace

First, register the marketplace with your AI assistant to enable plugin discovery.

claude plugin marketplace add phuryn/pm-skills

Source: README.md § Installation – Claude Code

Install the Plugin Under Test

Each domain (toolkit, discovery, strategy) lives in its own plugin folder. Install the specific plugin you want to validate.

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

Execute Slash Commands

Run a command to trigger the skill chain. This acts as your integration test by forcing the assistant to load the relevant skill files and execute the logic.

/discover "AI-powered meeting summarizer for remote teams"

When executed, the assistant will:

• Brainstorm ideas using pm-product-discovery/skills/brainstorm-ideas-new/SKILL.md
• Identify risky assumptions
• Prioritize those assumptions using the Impact × Risk matrix defined in pm-product-discovery/skills/prioritize-assumptions/SKILL.md
• Propose experiments

Source: pm-product-discovery/commands/discover.md

Validate the Output

Verify that the response contains expected sections: a structured list of ideas, an Impact × Risk matrix, and experiment suggestions. This manual verification serves as the "test assertion" for the skill's correctness.

Alternative Testing via OpenCode

For OpenCode users, test the skills by copying the skill files directly into your project and invoking them through the OpenCode interface.

# Copy skill files into the OpenCode project

for p in pm-*/; do cp -r "$p/skills/"* .opencode/skills/; done

# Invoke via OpenCode (exact command format depends on your UI)

opencode run skill=brainstorm-ideas-new "AI-powered meeting summarizer"

Key Files for Testing Workflows

Understanding these source files helps you verify what should happen during testing:

• README.md – Contains high-level installation and usage instructions for Claude Code integration.
• CLAUDE.md – Defines the marketplace format and specifies how skills and commands are loaded by Claude.
• pm-product-discovery/commands/discover.md – Defines the /discover command structure, which serves as the primary integration test workflow.
• pm-product-discovery/skills/brainstorm-ideas-new/SKILL.md – Example skill executed during /discover testing.
• pm-product-discovery/skills/prioritize-assumptions/SKILL.md – Skill that generates the Impact × Risk matrix used to validate test output.

Summary

• The phuryn/pm-skills repository contains no pytest or unittest framework; traditional unit tests do not exist.
• Testing is performed via integration testing by installing plugins and executing slash commands in an AI assistant environment.
• Use claude plugin marketplace add phuryn/pm-skills followed by claude plugin install to set up the test environment.
• Execute commands like /discover and manually validate the structured output against expected product management workflows.
• Key skill definitions reside in SKILL.md files within plugin directories such as pm-product-discovery/skills/.

Frequently Asked Questions

Does phuryn/pm-skills use pytest or unittest?

No. The repository contains no test_*.py modules, no test configuration in setup.cfg or pyproject.toml, and no CI workflows for automated testing. It uses markdown-based skill definitions that are tested through manual integration in AI assistant environments.

How do I test a specific skill in the repository?

Install the plugin containing the skill using claude plugin install <plugin-name>@pm-skills, then invoke the slash command that triggers that skill chain. For example, test the prioritization skill by running /discover and verifying the Impact × Risk matrix appears in the output, as defined in pm-product-discovery/skills/prioritize-assumptions/SKILL.md.

Can I automate testing for pm-skills plugins?

While the repository provides no built-in automation, you can script a programmatic check by calling the Claude Code API (or your assistant's API) with command text and inspecting the returned JSON for expected sections. You must write your own wrapper script, as the repository does not provide testing utilities.

What validates that the skills work correctly?

Validation occurs through manual inspection of the AI assistant's response. When you run a command like /discover, the assistant loads skills from paths such as pm-product-discovery/skills/brainstorm-ideas-new/SKILL.md and returns structured output. Correctness is determined by checking that the output contains expected components: idea lists, risk matrices, and experiment proposals.

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