creating-effective-skills
Creating Effective Skills
Guide for creating agent skills that follow Claude's official best practices.
Core Principles
Concise is Key: The context window is shared. Only add what Claude doesn't already know. Default assumption: Claude is already very smart.
Progressive Disclosure: Skills load in three levels:
- Metadata (~100 tokens) - always loaded
- SKILL.md body (<5k tokens) - when triggered
- Bundled resources - as needed
Keep SKILL.md small: Target ~200 lines, maximum 500 lines. Move detailed content to reference files aggressively.
Single Responsibility: Each skill does one thing well.
Workflow
More from taisukeoe/agentic-ai-skills-creator
improving-skills
Improve existing agent skills based on user feedback and best practices. Use when the user wants to fix, enhance, or refactor an existing skill. Gathers user feedback first, then applies technical analysis and implements improvements.
30reviewing-skills
Review skill files for best practices compliance (naming, description, structure, size). Use when checking SKILL.md quality or getting feedback before publishing. Static analysis only - does NOT execute the skill.
23running-skills-edd-cycle
Guides evaluation-driven development (EDD) process for agent skills. Use when setting up skill testing workflows, creating skill evaluation scenarios, or establishing Claude A/B feedback loops for skill validation. Provides development methodology, not content guidance.
17setting-up-devcontainers
Generate devcontainer configurations for Claude Code development environments. Use when setting up development containers with Claude Code and optional Codex CLI. Automatically detects marketplace.json for plugin marketplace configurations.
15reviewing-plugin-marketplace
Review Claude Code plugin marketplace configurations against official best practices. Use when analyzing marketplace.json and plugin.json files for structural issues, common errors, path validation, and consistency with Anthropic's official format. Detects repository URL mismatches, incorrect source paths, and missing required fields.
13evaluating-skills-with-models
Evaluate skills by executing them across sonnet, opus, and haiku models using sub-agents. Use when testing if a skill works correctly, comparing model performance, or finding the cheapest compatible model. Returns numeric scores (0-100) to differentiate model capabilities.
12