Design, test, and iterate Claude skills through a structured draft-to-evaluation workflow.
npx skills add https://github.com/Ikki-Group/erp --skill skill-creator-ikki-group
Please help me install this Agent Skill. Skill: skill-creator Source: https://github.com/Ikki-Group/erp/tree/main/.agents/skills/skill-creator Command: npx skills add https://github.com/Ikki-Group/erp --skill skill-creator-ikki-group
The Skill Creator provides a structured approach to designing, testing, and refining Claude skills, enabling teams to systematically capture intent, test performance, and iterate to achieve reliable triggering and high-quality outputs.
Draft a new skill using the available SKILL.md scaffolding, then run the evaluation loop to test, improve, and benchmark it.
High-intent search queries and answers about installing and using this skill.
To create and iterate Claude skills, use a structured workflow that guides teams from drafting to evaluation and benchmarking, applying bundled scripts and references to ensure reliable triggering and high-quality outputs.
Running evals and benchmarking an existing skill requires reusable scaffolding and scripts that collect metrics, test performance, and generate reports to optimize triggering descriptions and overall quality.
Skill evaluation and testing works by coordinating bundled resources, using scripts for deterministic tasks and references for background, to support reproducible evals and repeatable skill development.
Designing and testing skills requires PyYAML as a dependency, utilizing the provided SKILL.md scaffolding and assets for output templating to run the evaluation loop effectively.
Automated skill iteration is not suitable when deterministic tooling and reproducible evals are unnecessary, or when a project lacks the structured workflow needed to capture intent and test performance.
emellebyDesign, document, and package Claude skills with SKILL.md frontmatter.
timurgaleevMonitor post-deploy web apps for defects and performance regressions.
chicanoandres702Evaluate LLM outputs with automated metrics, human judgments, and A/B testing.
dotnetAutomate ad hoc performance benchmarking for dotnet/runtime code changes.
