search-router

To optimize token usage during code search, you need a routing mechanism that selects the most token-efficient tool based on query type. This approach prioritizes TLDR Search and AST-grep over Grep and Read, achieving up to 95% token savings for code exploration queries.

The best way to search structural code patterns is using AST-grep for structural analysis. By routing structural queries to AST-grep instead of standard text matching tools, you get precise pattern detection while minimizing unnecessary token consumption.

You need a search router for code exploration when queries involve semantic understanding or finding usages like a function call. The router intercepts these code-related queries and redirects them to TLDR Search, avoiding the significantly higher token cost of standard Grep.

To find where a class is defined efficiently, route your code exploration query to TLDR Search. This intelligent tool selection prevents unnecessary token consumption by matching the query nature with the most efficient available search mechanism.

Yes, the search router handles literal text matching differently from semantic code understanding by evaluating the query type. It separates code exploration, structural analysis, semantic understanding, and literal text matching to choose between TLDR Search, AST-grep, Grep, or Read.