Claude Code Skills System Reference FAQs

Question 1

How do I inject dynamic data into a Claude Code skill?

Accepted Answer

You can use the '!command' syntax within your SKILL.md file. This executes a shell command and replaces the placeholder with the command's output before the prompt is sent to Claude.

Question 2

Why is my skill not being automatically invoked by Claude?

Accepted Answer

This usually happens if your total skill metadata exceeds the ~15,000 character limit, causing truncation. You can still invoke truncated skills manually using the /skill-name command.

Question 3

How do I fix the YAML multiline bug in skill descriptions?

Accepted Answer

Avoid using YAML multiline indicators like '>' or '|'. Instead, use a single-line string or wrap the description in single quotes to ensure the Claude Code indexer parses it correctly.

Question 4

What is 'context: fork' and when should I use it?

Accepted Answer

Setting 'context: fork' runs the skill in an isolated subagent context without access to the main conversation history. It is best used for specific tasks that require a clean slate or a specific agent type like 'Explore' or 'Plan'.

Question 5

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

Accepted Answer

As of January 2026, skills and slash commands are unified. A file located in .claude/skills/ or .claude/commands/ functions identically, though the skills directory is recommended for advanced features like hooks.

Claude Code Skills System Reference

Claude Code Skills System Reference

Acerca de

Características Principales

Casos de Uso

Acerca de

Características Principales

Casos de Uso