skill-creator

To create an AI agent skill, structure your project with a SKILL.md file and resource directories like scripts/ and references/. Follow a progressive disclosure approach, outlining requirements, building components, and iterating on the design.

The correct anatomy for structuring a custom AI skill includes a core SKILL.md file alongside bundled resource directories such as scripts/, references/, and assets/. This modular structure ensures the skill remains self-contained and effectively extends agent capabilities.

Best practices for writing a SKILL.md file emphasize conciseness, progressive disclosure, and defining clear degrees of freedom. This ensures the AI agent understands the skill's scope and utilizes bundled resources effectively without being overwhelmed by unnecessary details.

When developing AI skills, bundle scripts and reference documents by organizing them into dedicated directories like scripts/ and references/ within the skill unit. This modular organization keeps custom tools and API integrations self-contained and easily accessible to the agent.

Prior prompt engineering experience is beneficial but not strictly required to build modular AI skills. The skill creation process provides comprehensive guidance on core principles and step-by-step development, helping developers structure capabilities regardless of their starting expertise.

You should avoid using a modular skill structure when an AI agent capability requires tightly coupled, monolithic logic that cannot be cleanly separated into scripts and references. If progressive disclosure restricts necessary context, a simpler direct prompt approach may be more effective.