Claude Code Skills Reference FAQs

Question 1

How does Claude decide which skill to use automatically?

Accepted Answer

Claude uses the 'description' field in the skill's frontmatter to determine relevance. If the description matches the user's current task or intent, the skill is automatically loaded into the context.

Question 2

What is the difference between a Claude Code skill and a slash command?

Accepted Answer

As of January 2026, the systems are unified. A file placed in the commands or skills directory creates a slash command, but the skill format is recommended for its support of advanced frontmatter and supplemental files.

Question 3

What is 'context: fork' in the skill configuration?

Accepted Answer

Setting context to 'fork' runs the skill in a subagent environment. This isolates the skill's operations from the main conversation history, which is useful for specialized tasks like research or planning.

Question 4

How can I inject live data into a skill prompt?

Accepted Answer

You can use dynamic context injection with backticks and an exclamation mark (e.g., !`git status`) to run shell commands that replace placeholders with real-time data before Claude sees the prompt.

Question 5

What are lifecycle hooks in Claude Code skills?

Accepted Answer

Hooks allow you to execute shell commands at specific points, such as before a tool is used (PreToolUse), after a tool is used (PostToolUse), or when the task stops, enabling automated linting or validation.

Claude Code Skills Reference

Claude Code Skills Reference

主要功能

使用场景

主要功能

使用场景