Browse And Evaluate

If you've ever installed a skill and immediately regretted the context window explosion, this is your guardrail. It walks you through the full browse, evaluate, preview, dry-run, install cycle with sane defaults like limiting fields to name, tier, workArea, and description. The best part is the enforcement: it won't let you skip the dry-run step or batch-install more than three skills without confirmation. The preview command even sanitizes potential prompt injection, which is both clever and slightly paranoid in a good way. Treat this as the disciplined workflow you know you should follow but usually skip when you're in a hurry.

Install to Claude Code

npx -y skills add moizibnyousaf/ai-agent-skills --skill browse-and-evaluate --agent claude-code