Draft, evaluate, and iterate on AI skills with structured evals.
npx skills add https://github.com/marinvch/ai-os --skill skill-creator-marinvch
Please help me install this Agent Skill. Skill: skill-creator Source: https://github.com/marinvch/ai-os/tree/main/skill-creator Command: npx skills add https://github.com/marinvch/ai-os --skill skill-creator-marinvch
This Skill enables teams and individuals to create new skills, iteratively improve existing ones, and measure performance with structured evals and benchmarking, reducing iteration time and increasing reliability.
Use the run_loop workflow to draft, evaluate, and iterate on this Skill until it meets your needs.
High-intent search queries and answers about installing and using this skill.
To improve AI skill quality, run automated evaluation loops that spawn baseline and with-skill runs, collect performance metrics, and refine descriptions or test prompts across multiple iterations.
Packaging an AI skill involves drafting the SKILL.md frontmatter and body, bundling optional resources, and applying structured evaluations to ensure reproducible guardrails and measurement across iterations.
Yes, you can benchmark skill performance across different configurations to guide optimization and decision-making by comparing outcomes from automated evaluation loops.
No, this skill creation workflow requires no external dependencies, allowing teams and individuals to draft, evaluate, and iterate on capabilities immediately without environment setup barriers.
The best way to reduce iteration time is using the run_loop workflow to generate improved descriptions, test prompts against baselines, and apply feedback through structured automated evaluations.
coreyjaDraft SKILL.md frontmatter and structure guidance for Claude Code skills.
timurgaleevMonitor post-deploy web apps for defects and performance regressions.
npc-chrisAutomate website performance benchmarking with page load times and Core Web Vitals.
cogni-daoDraft and refine Claude skill descriptions with evaluation loops and benchmarking.
