What Is the Difference Between Skills and Commands in PM Skills?

In PM Skills, skills are atomic knowledge units that load automatically when relevant, while commands are user-triggered workflows that orchestrate one or more skills into complete end-to-end processes.

The phuryn/pm-skills repository organizes product management expertise into a structured system built around two core abstractions: skills and commands. Understanding the difference between skills and commands in PM Skills is essential for effectively leveraging the toolkit, whether you are forcing a specific expertise load or executing a comprehensive workflow.

Skills: Atomic Knowledge Units

Skills serve as the foundational building blocks of the PM Skills ecosystem. Each skill encodes a discrete product-management framework, analysis method, or writing guideline as a standalone knowledge unit.

According to the source code, skills are loaded automatically whenever Claude (or another AI assistant) detects that the conversation content would benefit from that specific expertise. This automatic loading requires no explicit user invocation.

You can force a skill to load using a prefix syntax such as /plugin-name:skill-name or simply /skill-name.

The detailed instructions for a specific skill reside in dedicated files within the repository. For example, pm-toolkit/skills/review-resume/SKILL.md contains the exact guidelines the AI follows when reviewing a product manager resume.

Commands: User-Triggered Workflows

Commands represent the orchestration layer of the architecture. Unlike skills, which provide passive expertise, commands are active workflows that users trigger explicitly with a slash (/) prefix.

A command runs a scripted sequence that typically:

• Collects user input
• Calls one or more relevant skills
• Formats the output
• Suggests follow-up actions

For instance, the /review-resume command parses a resume input, invokes the review-resume skill, and produces a structured scorecard with improvement suggestions. This logic is defined in pm-toolkit/commands/review-resume.md.

Complex commands chain multiple skills together. The /discover command executes a full product discovery workflow by sequentially running the brainstorm-ideas, identify-assumptions, prioritize-assumptions, and brainstorm-experiments skills.

Key Architectural Differences

The distinction between these components can be summarized across three dimensions:

• Invocation: Skills load automatically based on context (or via forced prefix), while commands require explicit user initiation with /
• Scope: Skills contain focused domain knowledge; commands assemble end-to-end processes
• Composition: Skills are atomic; commands are composite, often orchestrating multiple skills

How to Invoke Skills and Commands

Understanding the syntax differences helps you choose the right tool for your needs.

To force-load a skill directly:

User: /review-resume "Here is my resume..."
Claude: (loads the `review-resume` skill automatically and returns the review)

To execute a command that chains multiple skills:

User: /discover AI-powered meeting summarizer
Claude: Executes the `brainstorm-ideas`, `identify-assumptions`, `prioritize-assumptions`, 
and `brainstorm-experiments` skills in sequence, then returns the full discovery output

The first example demonstrates a skill being called implicitly (or forced with a prefix). The second shows the /discover command internally invoking multiple skills to complete a comprehensive product-discovery workflow.

Summary

• Skills are atomic knowledge units that encode specific product management expertise, stored in files like pm-toolkit/skills/review-resume/SKILL.md
• Commands are user-triggered workflows defined in files like pm-toolkit/commands/review-resume.md that orchestrate skills into end-to-end processes
• Skills load automatically based on conversation context or via forced prefixes, while commands require explicit / invocation
• Commands can chain multiple skills together, such as /discover running four distinct skills in sequence

Frequently Asked Questions

Can I force a skill to load without using a command?

Yes. While skills typically load automatically when the AI detects relevant context, you can force a skill to load using the prefix syntax /plugin-name:skill-name or simply /skill-name according to the repository's README.md documentation.

How does a command know which skills to use?

Commands are scripted workflows that explicitly reference specific skills by name. For example, the /review-resume command is configured to call the review-resume skill, as defined in pm-toolkit/commands/review-resume.md. Complex commands like /discover contain hardcoded sequences that invoke multiple skills in a specific order to complete the workflow.

Are plugins different from skills and commands?

According to the README.md section on architecture, PM Skills is organized around three core concepts: Skills, Commands, and Plugins. While skills provide knowledge and commands provide workflows, plugins represent the third organizational layer that likely encapsulates related skills and commands into functional groups.

Where are skill definitions stored in the repository?

Skill definitions reside in the pm-toolkit/skills/ directory. Each skill has its own subdirectory containing a SKILL.md file that defines the specific instructions and knowledge. For example, the resume review skill instructions are located at pm-toolkit/skills/review-resume/SKILL.md.

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