skill-creator

To create and test Butler skills using Python, use a structured development workflow with standard directories like SKILL.md and scripts. You can run bulk test cases against a baseline and measure token consumption to ensure quality before deployment.

The standard structure for Butler skill development organizes files into SKILL.md, scripts, and references directories. This layout ensures consistent and efficient development, automated testing, and visual evaluation of skill outputs.

Yes, you need PyYAML and webbrowser dependencies to develop Butler skills. The development process requires a Butler environment and Python with these specific dependencies installed to support skill creation, testing, and description optimization.

Yes, you can measure token consumption and execution time for Butler skills. Performance analysis tools automatically track these key metrics during testing, allowing you to optimize skill behavior and resource usage before deployment.

To improve trigger accuracy for skill descriptions, use iterative description optimization. This process refines the skill's metadata and parameters until the desired trigger accuracy and response quality are achieved during testing.

The best way to qualitatively assess Butler skill outputs is using the eval-viewer tool. It allows you to visually evaluate test results, compare outputs against baselines, and provide direct feedback to iterate on the skill design.