Compatible AI Agents for PM Skills Marketplace: Claude, OpenCode, and Beyond

The PM Skills Marketplace supports Claude Code, Claude Cowork, OpenCode, Gemini CLI, Cursor, and any markdown-aware AI agent through its universal skill format.

The phuryn/pm-skills repository is an open-source Claude plugin marketplace that bundles nine specialized product management plugins. Each plugin contains markdown-based Skills and optional Commands that enable AI agents to execute complex PM workflows. Because the skills follow a universal text-based schema, compatible AI agents for PM Skills Marketplace extend beyond the Claude ecosystem to any platform that can read markdown files.

Architecture of the PM Skills Marketplace

The repository is structured as a layered system that separates the marketplace registry from individual plugin implementations.

The Plugin Layer and Marketplace Registry

At the root level, .claude-plugin/marketplace.json serves as the central registry that lists all nine available plugins. This JSON file allows Claude Code, Claude Cowork, and compatible agents to discover and install the entire marketplace or select individual plugins.

Each plugin folder (such as pm-toolkit, pm-product-strategy, pm-product-discovery, pm-market-research, pm-data-analytics, pm-marketing-growth, pm-go-to-market, pm-execution, and pm-ai-shipping) contains its own plugin.json manifest under .claude-plugin/. These manifests describe the plugin's skills and commands, enabling automatic loading when relevant to a conversation.

Skills and Commands Structure

Skills are plain markdown files (SKILL.md) stored within each plugin's skills/ directory. They follow a universal schema including title, description, usage example, and output format. For example, pm-toolkit/skills/review-resume/SKILL.md demonstrates the standard format that any AI agent can parse.

Commands are Claude-specific workflow definitions stored in commands/ folders (e.g., pm-product-discovery/commands/discover.md). These chain multiple skills together into end-to-end workflows triggered by slash commands like /discover.

Compatible AI Agents and Installation Methods

The PM Skills Marketplace supports multiple installation patterns depending on your AI agent's capabilities.

Claude Code (CLI)

For users running Claude Code in the terminal, install the marketplace using the built-in plugin manager:

# Add the marketplace (run once)

claude plugin marketplace add phuryn/pm-skills

# Install individual plugins

claude plugin install pm-toolkit@pm-skills
claude plugin install pm-product-strategy@pm-skills
claude plugin install pm-product-discovery@pm-skills
claude plugin install pm-market-research@pm-skills
claude plugin install pm-data-analytics@pm-skills
claude plugin install pm-marketing-growth@pm-skills
claude plugin install pm-go-to-market@pm-skills
claude plugin install pm-execution@pm-skills
claude plugin install pm-ai-shipping@pm-skills

Claude Cowork (UI)

For the web-based Claude Cowork interface:

1. Open Customize → Browse plugins → Personal → +
2. Select Add marketplace from GitHub and enter phuryn/pm-skills
3. All nine plugins appear automatically for selective installation

OpenCode and Markdown-Aware Agents

For agents like OpenCode, Gemini CLI, Cursor, or Kiro that consume markdown knowledge files, use the copy-to-agent script provided in the repository:

# Copy all skill folders into an OpenCode project

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

This approach bypasses the Claude-specific plugin system and imports the raw skills directly into your agent's context directory.

Using Skills and Commands Across Agents

Once installed, consumption patterns vary by agent capability.

Direct Skill Invocation

Any markdown-aware AI agent can invoke skills by referencing the skill content. For example, asking:

What are the riskiest assumptions for my AI-driven meeting-summarizer idea?

This triggers the identify-assumptions-new skill from pm-product-discovery, returning a structured list of assumptions regardless of the AI platform.

Command Chaining (Claude Only)

Claude-specific Commands enable multi-step workflows unavailable in other agents. The /discover command demonstrates this by automatically chaining:

1. brainstorm-ideas-new
2. identify-assumptions-new
3. prioritize-assumptions
4. brainstorm-experiments-new

To execute this in Claude Code or Cowork, simply type:

/discover AI-powered meeting summarizer for remote teams

The result is a complete discovery workflow output ready for stakeholder review.

Key Files and Repository Structure

Understanding these core files helps developers extend the marketplace or integrate it with custom agents:

• .claude-plugin/marketplace.json – Central registry defining all nine plugins for Claude agents
• pm-*/.claude-plugin/plugin.json – Individual plugin manifests describing available skills and commands
• pm-toolkit/skills/review-resume/SKILL.md – Reference implementation of the universal skill format
• pm-product-discovery/commands/discover.md – Example of a multi-skill workflow command
• README.md – Installation instructions and feature matrix
• CONTRIBUTING.md – Guidelines for adding new skills or plugins
• LICENSE – MIT license permitting open use and modification

Summary

• The PM Skills Marketplace is a Claude plugin ecosystem compatible with any AI agent that reads markdown
• Nine plugins cover the full product management lifecycle from strategy to execution
• Claude Code and Cowork use native plugin installation via claude plugin marketplace add
• OpenCode, Gemini CLI, Cursor, and Kiro import skills using the copy-to-agent script that extracts markdown files from pm-*/skills/ directories
• Skills are universal markdown files auto-loaded when relevant; Commands are Claude-specific workflow chains
• Primary limitation: Command chaining (/discover, etc.) requires Claude; other agents use skills individually

Frequently Asked Questions

Can I use PM Skills Marketplace with ChatGPT or GPT-4?

While the repository is optimized for Claude, you can manually copy the markdown skill files into ChatGPT's custom instructions or knowledge base. The skills in pm-*/skills/*/SKILL.md are plain text that any GPT model can process, though you will not have access to the automated command chaining available in Claude.

What is the difference between a Skill and a Command in this marketplace?

A Skill is a standalone markdown module (e.g., identify-assumptions-new) that defines a specific PM task's methodology, inputs, and outputs. A Command is a Claude-specific orchestration file that chains multiple skills into a workflow (e.g., /discover running four skills sequentially). Skills work in any markdown-aware agent; Commands require Claude Code or Cowork.

How do I contribute a new skill to the marketplace?

Fork the phuryn/pm-skills repository and follow the structure in CONTRIBUTING.md. Create a new SKILL.md file in the appropriate plugin's skills/ directory following the universal schema (title, description, usage example, output format). Submit a pull request with your skill; if accepted, it will appear in the .claude-plugin/marketplace.json registry for all compatible agents.

Is the PM Skills Marketplace free to use for commercial projects?

Yes. The repository is licensed under MIT, as documented in the LICENSE file. You can install the skills in commercial Claude workspaces, copy them to corporate OpenCode instances, or integrate them into proprietary AI agent setups without restriction.

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