How to Force Load a Specific PM Skill in the PM Skills Marketplace

Use the slash command syntax /skill-name or /plugin-name:skill-name to force a specific PM skill to load, ensuring it takes precedence over general knowledge in Claude-compatible interfaces.

The PM Skills Marketplace typically loads skills automatically based on conversation context. When you need explicit control over which framework or methodology drives the response, you can force load a specific PM skill using dedicated slash commands. This guide explains the implementation details based on the phuryn/pm-skills repository source code.

Syntax Options for Forcing Skill Loading

The repository supports two distinct syntax patterns for forcing skill activation, as documented in README.md at line 40.

Short-Hand Syntax

Use /skill-name when the skill identifier is unique across the entire marketplace. This approach works when no other plugin defines a skill with the same name, offering the quickest path to activation.

Fully Qualified Syntax

Use /plugin-name:skill-name to eliminate ambiguity when multiple plugins contain skills with identical identifiers. This explicit referencing ensures you load the exact skill from the intended plugin, which is critical in a marketplace with overlapping functionality.

Practical Implementation Examples

You can execute these commands in Claude Code, Claude Cowork, or any assistant interface that reads the PM Skills Marketplace. Below are concrete implementations based on the pm-toolkit plugin structure.

Force loading a skill with the short-hand command:

/review-resume

Explicitly referencing the plugin to avoid naming collisions:

/pm-toolkit:review-resume

Using the syntax in a command file (e.g., commands/review-resume.md) to chain skills:

/suffix: /pm-toolkit:review-resume

When executed, the assistant loads the skill definition from pm-toolkit/skills/review-resume/SKILL.md and applies its structured response format. This process overrides generic knowledge that might otherwise dominate the answer.

How Forced Loading Works Under the Hood

When you issue a force load command, the system prepends the skill to the knowledge base for that specific exchange. The implementation relies on specific file structures within the repository:

• Plugin Metadata: The pm-toolkit/.claude-plugin/plugin.json file defines the plugin name (pm-toolkit) used in the fully qualified syntax.
• Skill Definition: The actual skill content resides in pm-toolkit/skills/review-resume/SKILL.md, which provides the structured guidance (introduction, feedback on best practices, conclusion) that dictates the model's output format.

This mechanism guarantees that the skill's framework-specific guidance drives the response rather than general conversational knowledge.

Summary

• Force load specific PM skills using /skill-name for unique identifiers or /plugin-name:skill-name for explicit plugin targeting.
• The syntax functions across all Claude-compatible interfaces, including Claude Code and Claude Cowork.
• Skills are loaded from their respective SKILL.md files within the plugin's skills/ directory.
• Forced loading prepends the skill to the conversation's knowledge base, ensuring framework-driven outputs.

Frequently Asked Questions

What is the difference between /skill-name and /plugin-name:skill-name?

The short-hand /skill-name works when the skill name is unique across the marketplace, while /plugin-name:skill-name explicitly targets a skill within a specific plugin. Use the fully qualified syntax to avoid collisions when multiple plugins define skills with identical names.

Where are skill definitions stored in the repository?

Skill definitions are stored in SKILL.md files within plugin-specific directories, such as pm-toolkit/skills/review-resume/SKILL.md. The plugin metadata in .claude-plugin/plugin.json identifies the plugin name used in forced loading commands.

Can I force load multiple skills in a single conversation?

Yes, you can issue multiple slash commands to load several skills simultaneously. Each forced skill prepends its structured guidance to the context, allowing you to combine frameworks from different plugins.

Does forcing a skill override the assistant's general knowledge?

Yes, when you force load a skill, it prepends the structured guidance to the knowledge base for that exchange. This ensures the skill's specific framework drives the response, taking precedence over general knowledge that might otherwise dominate the output.

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