code-discovery

You can map natural language queries to code symbols by using static analysis to generate a symbol_index.json manifest. This ranks exported symbols against user goals, providing file paths and signatures for large codebases.

Deterministic symbol ranking assigns consistent scores to code symbols matched from natural language queries, ensuring identical search results every time. It simplifies tooling integration by providing predictable symbol discovery and routing.

You generate a symbol index by running static analysis over a repository to extract and rank exported symbols. This creates a symbol_index.json manifest containing file paths and signatures for subsequent natural language query routing.

Yes, you can retrieve dependency graphs for specific symbols using the depGraph function. It analyzes a given symbol and returns its dependency relationships, helping you map connections within the codebase structure.

Yes, code discovery is specifically designed for large codebases. It uses static analysis to map natural language goals to concrete exported symbols, generating a symbol index that scales for extensive repositories and consistent tooling integration.

A limitation of natural language to code symbol mapping is its reliance on static analysis, meaning it only targets exported symbols and cannot capture dynamic runtime behaviors. It generates deterministic manifests rather than executing code paths.