The Difference Between Skills and Commands in the PM Skills Marketplace

Skills are self-contained expertise modules that provide reusable product management knowledge, while commands are slash-triggered orchestration workflows that chain multiple skills together to solve end-to-end problems.

The phuryn/pm-skills repository implements a modular architecture that separates discrete domain knowledge from complex multi-step workflows. Understanding how these two components interact allows you to leverage the marketplace effectively, whether you are invoking individual capabilities or running comprehensive product management processes.

What Are Skills?

A skill is a self-contained unit of domain knowledge or a reusable workflow that encodes a specific product-management framework. Skills include templates like value-proposition canvases, analysis methods such as SWOT, or guided workflows like resume reviews. According to the repository's README (lines 32-44), skills are designed to be loaded automatically when the model detects relevant context, though they can also be invoked directly by name without any prefix.

Skills are defined in individual directories containing SKILL.md files. For example, the resume review capability resides in pm-toolkit/skills/review-resume/SKILL.md. This file contains the structured prompts and logic necessary to perform a PM resume review, making the expertise portable across different commands and contexts.

review-resume

When you type review-resume without a leading slash, the model pulls the skill definition and returns feedback immediately. This implicit invocation allows skills to act as organic extensions of the model's knowledge base.

What Are Commands?

A command is a user-triggered, higher-level workflow that strings together one or more skills to solve a complete problem. Commands are always explicit, requiring a leading slash (/), and they drive multi-step interactions that may involve state management across several skill invocations.

Command definitions are stored as markdown files in commands/ directories within each plugin. For instance, pm-toolkit/commands/review-resume.md implements the /review-resume command, which wraps the underlying skill with additional orchestration logic. More complex examples include pm-product-discovery/commands/discover.md and pm-go-to-market/commands/plan-launch.md, which coordinate multiple skills into structured pipelines.

/discover AI-powered meeting summarizer for remote teams

When executed, the /discover command runs a discovery workflow that internally chains four distinct skills: brainstorm-ideas, identify-assumptions, prioritize-assumptions, and brainstorm-experiments (README lines 60-62).

Key Architectural Differences

The relationship between these components follows a clear separation of concerns:

• Purpose: Skills provide discrete, reusable expertise—single templates or analysis methods that can be shared across many workflows. Commands provide orchestration, composing complete processes from individual skills.

• Invocation: Skills are invoked implicitly when the model detects relevance or explicitly by name without prefixes (swot-analysis). Commands are always explicit via slash syntax (/plan-launch).

• Scope: A skill handles a single, focused task. A command manages a composite, multi-step task that may invoke several skills sequentially or conditionally.

• Discovery: In the marketplace UI, skills appear under each plugin's "Skills" section, while commands appear under "Commands."

How Commands Chain Skills

The orchestration layer in commands demonstrates the power of this architecture. When you invoke /plan-launch for an AI code review tool, the command defined in pm-go-to-market/commands/plan-launch.md coordinates multiple strategic skills including gtm-strategy and ideal-customer-profile to generate a comprehensive launch plan (README lines 55-57).

This composition happens through the plugin manifest system. The file pm-toolkit/.claude-plugin/plugin.json registers both skill and command files, making them discoverable to the orchestration engine while maintaining their distinct roles.

Summary

• Skills are atomic expertise units stored in SKILL.md files, invoked by name or context, and handle single tasks like review-resume or swot-analysis.
• Commands are orchestration workflows stored in commands/*.md, invoked with a leading /, and chain multiple skills to solve complex problems like /discover or /plan-launch.
• The architecture separates reusable knowledge from workflow logic, allowing skills to be shared across different commands while maintaining clear boundaries between simple invocation and complex process management.
• File locations follow strict conventions: skills reside in skills/<name>/SKILL.md while commands reside in commands/<name>.md.

Frequently Asked Questions

Can I invoke a skill without using a command?

Yes. Skills can be invoked directly by typing their name without a leading slash, such as review-resume or swot-analysis. The model detects these references and loads the appropriate SKILL.md file from the corresponding plugin directory automatically.

What's the difference between /review-resume and the review-resume skill?

The review-resume skill defined in pm-toolkit/skills/review-resume/SKILL.md contains the core logic and prompts for analyzing PM resumes. The /review-resume command defined in pm-toolkit/commands/review-resume.md wraps this skill in a command interface, potentially adding pre-processing, parameter validation, or post-processing steps that prepare the input before invoking the underlying skill.

How does the /discover command use multiple skills?

The /discover command implements a pipeline that sequentially invokes four skills: brainstorm-ideas, identify-assumptions, prioritize-assumptions, and brainstorm-experiments. As documented in the README (lines 60-62), this command runs the complete "brainstorm → identify → prioritize → design experiments" discovery process by chaining these individual skills together in pm-product-discovery/commands/discover.md.

Where are skills and commands registered in the repository?

Both components are registered in the plugin manifest file located at pm-toolkit/.claude-plugin/plugin.json. This JSON configuration maps skill directories and command files to the plugin system, enabling the marketplace to discover and load both atomic skills and composite commands while maintaining their distinct architectural roles.

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