Skill Coach

소개

Skill Coach empowers developers to build expert-level Claude Code skills that transcend surface-level instructions by capturing deep domain knowledge known as shibboleths. It provides a structured framework for designing precise activation triggers, surfacing common pitfalls through anti-pattern detection, and managing complex information via progressive disclosure architecture. By focusing on decision trees rather than static templates, it ensures skills are both highly effective and token-efficient, helping users avoid common mistakes like vague descriptions or bloated instruction sets.

주요 기능

• Guides progressive disclosure implementation to maintain token efficiency
• Encodes domain-specific shibboleths and expert decision-making logic
• Provides structured templates for precise skill activation and NOT clauses
• Detects and prevents common anti-patterns in skill architecture
• Captures temporal knowledge to distinguish between legacy and modern practices
• 3 GitHub stars

사용 사례

• Creating new domain-specific Agent Skills from scratch
• Auditing and improving the activation rates of existing skills
• Refactoring large, vague skills into focused, composable expert modules