Skill Summary

Claude Skills are Agent Skills — they follow the Agent Skills open format. Skills are specialized workflows trained to work in a very specific way.

Claude Code extends the standard with additional features like: invocation control, subagent execution, dynamic context injection.

Claude Code has built-in skills already available in every session. Built-in skills will not show up under /skills — that only lists user-defined skills.

Built-in skills:

/simplify — reviews your recently changed files for code reuse, quality, and efficiency issues, then fixes them.
/batch — orchestrates large-scale changes across a codebase in parallel.
/debug — troubleshoots your current Claude Code session by reading the session debug log.
/loop — runs a prompt repeatedly on an interval while the session stays open.
/claude-api — loads Claude API reference material for your project's language.

Skills you create can be invoked automatically ("How does this code work?" triggers a matching skill) or directly (e.g. /explain-code src/auth/login.ts).

Where skills are located:

Enterprise — based on Claude Code settings.
Personal — ~/.claude/skills/<skill-name>/SKILL.md
Project — .claude/skills/<skill-name>/SKILL.md
Plugin — <plugin>/skills/<skill-name>/SKILL.md

Kinds of skills:

Reference content — adds knowledge.
Task content — does something.