Open Skills Logo
Open Skills

skill-developer

Create and manage Claude Code skills following Anthropic best practices. Use when creating new skills, modifying skill-rules.json, understanding trigger patterns, working with hooks, debugging skill activation, or implementing progressive disclosure. Covers skill structure, YAML frontmatter, trigger types (keywords, intent patterns, file paths, content patterns), enforcement levels (block, suggest, warn), hook mechanisms (UserPromptSubmit, PreToolUse), session tracking, and the 500-line rule.

Author

@sickn33

Category

AI Skill Development

Install

Hot:5

Download and extract to your skills directory

Copy command and send to OpenClaw for auto-install:

Download and install this skill https://openskills.cc/api/download?slug=sickn33-skills-skill-developer&locale=en&source=copy

Skill Developer - Complete Guide to Claude Code Skill Development

Skill Overview

Skill Developer is the authoritative guide for creating and managing custom skills for Claude Code. It helps developers master Anthropic’s official best practices, including skill architecture design, trigger configuration, the Hook mechanism, and the 500-line rule.

Use Cases

  • Create a new Claude Code skill

    • When you need to build a custom skill for a specific domain or task, this guide provides a complete workflow—from writing SKILL.md to configuring skill-rules.json—so that your skill conforms to Anthropic’s official standards.

  • Debug skill activation issues

    • When the skill you wrote is not being auto-activated, this guide offers a systematic debugging approach, including trigger testing, Hook validation, and troubleshooting common problems.

  • Implement progressive disclosure mode

    • When your skill documentation exceeds 500 lines, this guide teaches you how to keep the core SKILL.md concise using reference files and a layered structure, while still providing detailed reference documentation.

    Core Features

  • Dual Hook architecture system

    • Explains the operation mechanisms of two Hook types—UserPromptSubmit (proactive suggestions) and Stop Hook (gentle reminders)—to help developers understand how and when a skill is activated and called.

  • Full trigger type support

    • Covers four trigger methods: keyword triggers, intent modes (regular expressions), file paths (Glob), and content detection. Includes best-practice configuration examples for real scenarios.

  • Flexible execution-level control

    • Supports three execution levels—BLOCK (forcefully block), SUGGEST (suggest/remind), and WARN (warn)—allowing developers to choose the appropriate level of strictness based on the skill’s importance.

    Frequently Asked Questions

    How do I create a Claude Code custom skill?

    Creating a skill consists of five steps:

  • Create a SKILL.md file under .claude/skills/{skill-name}/ containing YAML frontmatter (name and description) and the skill content

  • Add the skill configuration to .claude/skills/skill-rules.json, defining the type, execution level, and trigger conditions

  • Use the command line to test whether the triggers work properly

  • Optimize keyword and pattern matching based on test results

  • Ensure SKILL.md is no more than 500 lines; anything beyond that goes into reference files

    • Why isn’t my skill being activated?

    A skill usually fails to activate for the following reasons:

  • Keywords don’t match: Check that description and promptTriggers.keywords include words the user is likely to input

  • Intent pattern is too strict: Complex regular expressions may not match how users actually phrase things

  • File path is wrong: The Glob paths in filePatterns don’t match the actual file locations

  • Hook isn’t registered: Confirm that the Hook file is correctly registered in .claude/settings.json

  • Session is skipped: A skill that has already been triggered in the same session won’t remind again (this is normal behavior)

    • It’s recommended to use npx tsx .claude/hooks/skill-activation-prompt.ts for manual testing.

    What’s the difference between Guardrail skills and Domain skills?

    The main differences are purpose and strictness:

    FeatureGuardrail skillDomain skill
    Typeguardraildomain
    Execution levelblocksuggest
    Prioritycritical/highhigh/medium
    PurposePrevent critical mistakes (e.g., wrong database column names)Provide domain guidance and best practices
    BehaviorPhysically blocks editing until the skill is usedSuggest using the relevant skill at the right time
    Typical scenariosdatabase-verificationbackend-dev-guidelines

    If your skill is meant to prevent runtime errors or data-safety issues, use Guardrail; if it’s meant to provide development guidance, use Domain.