Optimizes Claude's code search strategy to minimize token consumption and improve query accuracy.
The Maestro Search Hierarchy skill implements a token-efficient decision tree for code exploration, guiding Claude to use the most appropriate tool for every query. By prioritizing AST-grep for structural patterns, LEANN for semantic concepts, and standard Grep only for literal identifiers, it drastically reduces unnecessary context loading. This skill includes built-in enforcement hooks that prevent expensive 'Read' operations before relevant files are identified, ensuring faster responses and significantly lower API costs during complex development tasks.
Key Features
01Automatic hook enforcement via grep-to-leann scripts
02Semantic search via LEANN for conceptual discovery
03Token-efficient decision tree for code navigation
04AST-grep integration for structural code patterns
050 GitHub stars
06Significant reduction in context window usage
Use Cases
01Finding specific function or class definitions across large repositories
02Locating exact string identifiers or regex patterns efficiently
03Understanding high-level architecture and patterns without reading every file
