smart-explore

Search code structure across a repository using AST parsing by requesting ranked symbols and signatures to locate functions and classes before loading full file contents. This structural discovery maps where code elements live without guess-and-check reading.

Optimize tokens when navigating large code files by using a three-step workflow: search for a code map, outline the file skeleton, and unfold only the targeted symbol implementation. This avoids full-file reads and saves token usage.

Yes, tree-sitter AST parsing supports repository-wide symbol discovery across supported programming languages. It finds structural sections, functions, and classes to provide efficient navigation and targeted review across different codebases.

Extract a file skeleton by requesting a structural outline of the code file. AST parsing generates this skeleton by identifying code elements and structural sections, allowing you to understand the file architecture without loading the full content.

Structural code search works better than standard text search for code navigation because it uses tree-sitter AST parsing to find ranked symbols and structural sections. This targets code elements directly instead of matching arbitrary text strings.

Avoid reading full file contents during code exploration when you need to locate specific functions or classes in a large repository. Use structural search and symbol unfolding instead to inspect only relevant implementations and reduce token consumption.