Overview

Claude Spec Skill demonstrates all Claude Code specification frontmatter fields. It uses a minimal skill body to test parsing and handling of metadata such as name, description, argument-hint, disable-model-invocation, user-invocable, allowed-tools, model, context, agent, and hooks. This makes it a practical reference for validating spec compliance in development environments.

How This Skill Works

The skill's frontmatter defines metadata consumed by Claude Code during invocation. A PreToolUse Bash hook runs before tool usage, illustrating how hooks are specified and triggered. Certain fields, such as disable-model-invocation and user-invocable, enforce restricted interactions in practice.

When to Use It

• When validating parsing of all Claude Code spec fields in a new skill
• When testing hook execution via PreToolUse with a Bash matcher
• When verifying that model invocation is disabled for a read-only test
• When ensuring the user-invocable flag prevents manual invocation
• When checking allowed-tools syntax and effective gating (e.g., Read Write Bash(git:*))

Quick Start

1. Step 1: Inspect the SKILL.md frontmatter for name, description, and all fields.
2. Step 2: Trigger a test invocation to exercise PreToolUse and the Bash hook.
3. Step 3: Confirm that model invocation is disabled and the skill is not user-invocable.

Best Practices

• Review every frontmatter field for correct spelling and values
• Test each field's impact in a dry-run environment
• Verify hook definitions and their commands execute as expected
• Confirm the allowed-tools syntax supports patterns like git:*
• Document the purpose of each field for future spec readers

Example Use Cases

• CI suite validating parsing of Claude Code frontmatter fields
• Documentation sample showing all spec fields in SKILL.md
• QA test ensuring disable-model-invocation and user-invocable flags are honored
• Demo of PreToolUse Bash hook triggering a command
• Fork-context skill usage for reviewer agent scenarios

Frequently Asked Questions