Assess

When you need to answer "is this code good enough to ship?" this skill gives you structured evaluation across six dimensions: correctness, maintainability, security, performance, testability, and simplicity. It pulls from your project's memory server to compare against past decisions and conventions, then produces a 0-10 score with pros, cons, and actionable improvements. You can tune effort levels from quick assessments to an "xhigh" mode on Opus 4.8 that adds an uncertainty pass with per-dimension confidence scores. It's built for code reviews, architecture comparisons, and trade-off analysis. The handoff system lets you resume mid-assessment, and it respects model overrides and effort flags from the command line.

Install to Claude Code

npx -y skills add yonatangross/orchestkit --skill assess --agent claude-code