How to Force-Load Specific PM Skills Using the `/skill-name` Syntax

Use the /skill-name or /plugin-name:skill-name syntax to explicitly force-load a PM Skill regardless of conversation context, ensuring the skill's knowledge base is available immediately.

The phuryn/pm-skills repository organizes product management capabilities into modular skills that typically auto-load based on context triggers. However, you can override this behavior and force-load specific PM skills using slash syntax to guarantee a skill's framework or templates are active before you issue a prompt.

Understanding the Force-Load Syntax

PM Skills distinguish between skills (nouns that provide knowledge) and commands (verbs that execute actions). While skills auto-load when Claude detects relevant trigger phrases, you can force-load them using two syntax variants documented in the repository.

According to the public documentation in [README.md](https://github.com/phuryn/pm-skills/blob/main/README.md), the force-loading feature accepts both short and fully qualified forms (lines 41-42). The internal agent guidance in [CLAUDE.md](https://github.com/phuryn/pm-skills/blob/main/CLAUDE.md) repeats this rule for Claude agents (line 58), confirming the syntax is processed at the system level before skill execution.

Syntax Variations for Force-Loading

Short Form Syntax

The short form uses the forward slash followed directly by the skill identifier:

/skill-name

Claude internally resolves this to the appropriate plugin based on context or default mappings.

Fully Qualified Syntax

When multiple plugins might contain skills with similar names, or when you need explicit control, use the fully qualified form:

/plugin-name:skill-name

This syntax explicitly names the plugin namespace, preventing ambiguity and ensuring the correct skill loads.

Practical Code Examples

To force-load the review-resume skill from the pm-toolkit plugin, you can use either approach:

# Short form - relies on implicit plugin resolution

/review-resume

# Fully qualified form - explicit plugin declaration

/pm-toolkit:review-resume

After force-loading, follow up with your specific request:

Please evaluate my product-manager résumé against the XYZ+S best-practice checklist.

How Force-Loading Works Internally

When you issue a force-load command, Claude prepends the appropriate plugin prefix internally, ensuring the skill's front-matter and implementation are imported before processing your request. This operation guarantees that the skill's knowledge base—including frameworks, templates, or reference data—is available even if the conversation hasn't yet mentioned a related term.

Force-loading is particularly useful when you want a skill to take precedence over Claude's generic knowledge or when chaining multiple skills in a custom prompt. Because PM Skills are pure data (no executable code), this operation is safe and does not alter the runtime environment.

Key Files in the Repository

The force-loading mechanism is documented and implemented across these specific files:

File	Role	Location
README.md	Public documentation describing the force-loading syntax	README.md
CLAUDE.md	Agent-specific guidance repeating the syntax rules for Claude agents	CLAUDE.md
pm-toolkit/skills/review-resume/SKILL.md	Implementation of the review-resume skill showing the skill structure	SKILL.md
pm-toolkit/commands/review-resume.md	Command definition that chains the skill, demonstrating slash command usage	review-resume.md

Summary

• Force-loading bypasses auto-detection to activate a skill immediately using /skill-name or /plugin-name:skill-name syntax.
• The short form relies on implicit plugin resolution, while the fully qualified form explicitly targets a specific plugin namespace.
• Repository documentation in README.md (lines 41-42) and CLAUDE.md (line 58) formally defines this behavior.
• Force-loading ensures skill knowledge bases are available before prompt processing, useful for precedence control and skill chaining.
• PM Skills contain only data, making force-loading a safe operation that doesn't execute arbitrary code.

Frequently Asked Questions

What is the difference between auto-loading and force-loading a skill?

Auto-loading occurs when Claude detects trigger phrases in your conversation that match a skill's defined patterns, automatically importing the skill's context. Force-loading uses the /skill-name syntax to explicitly import a skill regardless of whether trigger phrases are present, ensuring immediate availability of the skill's knowledge base.

When should I use the fully qualified syntax versus the short form?

Use the fully qualified syntax (/plugin-name:skill-name) when multiple plugins might contain skills with identical names, or when you need absolute certainty about which skill loads. Use the short form (/skill-name) for convenience when the skill name is unique across your installed plugins or when the default resolution matches your intent.

Is force-loading a skill safe? Does it execute code?

Force-loading is completely safe. PM Skills are pure data structures containing frameworks, templates, and reference material—no executable code. The operation simply imports the skill's front-matter and knowledge base into the conversation context without altering the runtime environment or executing external programs.

Can I force-load multiple skills in a single prompt?

Yes, you can chain multiple force-load commands by including each /skill-name or /plugin-name:skill-name reference on separate lines at the beginning of your prompt. This loads all specified skills simultaneously, allowing you to combine knowledge bases from multiple PM Skills before issuing your request.

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