skill-creator

To create a SKILL.md file for Claude, define the skill's purpose, trigger contexts, expected outputs, dependencies, and interfaces so Claude can reliably trigger and execute it. This skill standardizes that creation process through interview-style elicitation and structured documentation.

The best way to evaluate Claude skills is to run automated trigger-eval loops that collect metrics and generate reports comparing skill versions. This supports continuous iteration until skill triggering becomes reliable across different contexts.

Interview-style elicitation works by guiding you through targeted questions to capture intent, gather requirements, and map them directly to the SKILL.md structure. This ensures the skill's purpose and trigger conditions are precisely documented before synthesis.

Yes, you need pyyaml installed to use this skill, as it is a required dependency for parsing configuration and running the automated evaluation and benchmarking loops that measure skill triggering reliability.

Yes, you can iterate on skill descriptions to improve trigger reliability by running evaluation loops, reviewing collected metrics from generated reports, and refining the SKILL.md prompts and references until the expected triggering behavior is consistently achieved.

The included components for synthesizing a skill are scripts for automation, references for mapping requirements, and assets for supporting the SKILL.md creation and evaluation process. These work together to produce a cohesive, testable skill entry.