code-parser skill
Reduce LLM token consumption when working with Dart, Python, and TypeScript codebases by indexing structure first, then reading only the lines that matter.
Core principle
Never read a full source file when you can read its structure first.
Full file read: ~15,000 tokens for a 2,000-line file
code-parser index: ~400 tokens for the same file
Targeted read: ~300 tokens per method body
Total saving: 70–97% fewer tokens per task
Workflow
Follow t
[Description truncada. Veja o README completo no GitHub.]