CLAUDE.md Linter

CLAUDE.md Linter grades AI project-instruction files across 10 checks — purpose clarity, role definition, prohibited behaviors, structure, output format, examples, context-budget awareness, tiered loading, memory rules, and agent routing — and returns a letter grade from A+ to F. Every failed check expands into a concrete recommendation, and a copy-to-clipboard badge makes the score shareable. Runs entirely in-browser with no backend; built on Base44, with a public link coming once published.