skill-creator

To structure an AI agent skill for progressive disclosure, you must organize bundled resources like scripts, references, and assets effectively while keeping the core SKILL.md concise. This ensures the AI agent loads details only when needed.

Creating an AI skill requires defining a SKILL.md file and organizing supporting components such as scripts, references, and assets. These bundled resources provide the execution logic and documentation for the AI agent.

Best practices for AI skill development include maintaining conciseness, applying progressive disclosure, and setting clear degrees of freedom. Properly structuring documentation and bundled resources enhances agent capabilities efficiently.

To integrate an external API into a new AI skill, you should structure the SKILL.md with instructions, organize the API documentation within the references directory, and include any necessary helper scripts in the scripts folder.

Use bundled scripts when your AI skill requires executable logic or automation, and use the references directory to store static documentation. This separation ensures the AI agent accesses code and context appropriately.

Building an AI skill using this guide does not require managing external dependencies, as the current implementation has none. You simply structure your internal scripts, references, and assets to function independently.