skill-developer

To create a Claude Code skill, you must define its structure using YAML frontmatter and configure trigger mechanisms like keywords, intent patterns, and file paths. This ensures the AI assistant activates the skill correctly based on user context.

Claude Code skills use UserPromptSubmit and PreToolUse hooks for activation and enforcement. These hooks allow skills to block, suggest, or warn against specific actions during AI-assisted development sessions.

You configure trigger conditions in skill-rules.json by defining keywords, intent patterns, file paths, and content patterns. These rules determine exactly when a specific skill activates during a coding session.

The 500-line rule is a best practice for Claude Code skill development that limits skill size to maintain clarity and effectiveness. It enforces progressive disclosure, keeping the AI assistant focused and manageable.

When building Claude Code skills, you can set enforcement levels to block, suggest, or warn. These levels control how strictly the AI assistant interrupts or advises the user during rule violations.

Yes, you can combine intent patterns, file paths, and content patterns to trigger a skill. This multi-signal approach ensures the AI assistant activates skills accurately based on both user prompts and active code context.