How the $ARGUMENTS Variable Works in Claude Code Skill Definitions

The $ARGUMENTS placeholder in SKILL.md files is replaced at runtime with the argument string a user provides after a slash command, enabling dynamic prompt injection without custom parsing code.

The anthropics/claude-plugins-community repository defines interactive capabilities for Claude Code through declarative skill definitions. The $ARGUMENTS variable provides a clean mechanism for passing free-form user input directly into skill prompts, eliminating the need for custom argument parsing logic.

Runtime Substitution Mechanism

When a user invokes a skill via slash command, the Claude Code platform captures everything after the command name as the argument string. The skill engine then performs template substitution before sending the prompt to the model.

Step-by-Step Argument Injection

1. User invokes the skill: /eli5 quantum computing

2. Claude Code extracts the argument string: quantum computing

3. Engine substitutes $ARGUMENTS with that string while building the prompt

4. The model receives the concrete prompt: Topic: quantum computing

This process occurs automatically for any SKILL.md file containing the $ARGUMENTS token.

Implementation in the Repository

According to the source code analysis, the $ARGUMENTS placeholder currently appears in a specific skill within the repository.

The eli5 Skill Definition

The only skill utilizing $ARGUMENTS is the eli5 skill, defined in eli5/skills/eli5/SKILL.md. This file contains the template:

Topic: $ARGUMENTS

When a user runs /eli5 photosynthesis, the engine transforms this template into Topic: photosynthesis before model inference.

Syntax Patterns and Examples

The $ARGUMENTS variable follows a consistent placement pattern within skill templates.

Basic Substitution Pattern

Skills typically frame user input with static context using this structure:

<some static text> $ARGUMENTS <more static text>

This pattern ensures the model receives properly contextualized instructions while incorporating dynamic user input.

Practical Runtime Examples

Below are concrete examples of how $ARGUMENTS resolves during execution:

Example 1: Simple Topic Assignment

• Template: Topic: $ARGUMENTS
• User command: /eli5 photosynthesis
• Result: Topic: photosynthesis

Example 2: Embedded Explanation Request

• Template: Explain $ARGUMENTS in simple terms.
• User command: /eli5 blockchain
• Result: Explain blockchain in simple terms.

Example 3: Step-by-Step Instructions

• Template: Give a step-by-step guide for $ARGUMENTS.
• User command: /eli5 making coffee
• Result: Give a step-by-step guide for making coffee.

Integration with Other Placeholders

While $ARGUMENTS handles slash command inputs, skills can combine this token with other placeholders such as $INPUT, $OUTPUT, and $CONTEXT to build richer, multi-dimensional prompts. This composability allows skills to reference conversation history, previous outputs, or structured inputs alongside the raw argument string provided by the user.

Summary

• $ARGUMENTS is a runtime placeholder in Claude Code SKILL.md files that gets replaced with user-provided text
• The variable appears specifically in eli5/skills/eli5/SKILL.md within the anthropics/claude-plugins-community repository
• Substitution happens automatically when users invoke skills with arguments (e.g., /eli5 quantum computing)
• No custom parsing code is required; the platform handles extraction and injection
• Skills can combine $ARGUMENTS with other placeholders like $INPUT and $CONTEXT for complex prompt engineering

Frequently Asked Questions

What is the $ARGUMENTS variable in Claude Code skills?

The $ARGUMENTS variable is a template placeholder used in SKILL.md files that automatically receives the text a user types after a slash command. It enables skills to accept free-form input without implementing custom argument parsing logic.

How does the runtime substitution of $ARGUMENTS work?

When a user invokes a skill like /eli5 quantum computing, Claude Code extracts the string quantum computing and replaces every instance of $ARGUMENTS in the skill's template with this value before sending the prompt to the model.

Where is $ARGUMENTS implemented in the anthropics/claude-plugins-community repository?

Currently, $ARGUMENTS is implemented only in the eli5 skill, specifically within the file eli5/skills/eli5/SKILL.md. This skill uses the pattern Topic: $ARGUMENTS to receive explanation topics from users.

Can $ARGUMENTS be used multiple times in a single skill template?

Yes, you can include $ARGUMENTS multiple times within a SKILL.md file, and each instance will be replaced with the same user-provided argument string during the substitution phase.

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