CortexAST

About

CortexAST revolutionizes AI coding by serving as an AI-native IDE backend, providing LLM agents with a 'God-tier' semantic understanding of code. Unlike traditional tools that flood context windows with noise, CortexAST leverages Tree-sitter (AST) and Rust to deliver precise, high-fidelity codebase insights. It drastically cuts token usage by focusing purely on architectural logic, enabling agents to navigate, read, find, and refactor code with unparalleled efficiency and accuracy, effectively transforming LLMs from mere 'slicers' to intelligent 'orchestrators' of code changes.

Key Features

• Precise extraction of specific AST nodes (functions/structs) from files (`read_symbol`)
• Hierarchical codebase mapping with exported symbols and smart filtering (`map_repo`)
• Semantic, AST-aware tracing of code usages, categorized by type (`find_usages`)
• Cross-language propagation checklists for microservice updates (`propagation_checklist`)
• AST snapshot comparisons for structural refactoring and semantic logic tracking (`Chronos`)
• 7 GitHub stars

Use Cases

• Enabling LLM agents to semantically explore and understand complex codebases.
• Optimizing token usage for AI code analysis and generation tasks.
• Facilitating accurate, structural refactoring and cross-service code propagation by AI.