tldr-code-analysis

To analyze an unfamiliar codebase without exceeding token limits, generate a token-efficient structural summary using AST, call graph, CFG, DFG, and PDG. This 5-layer analysis reduces token usage by up to 95% while preserving structural relationships for onboarding or refactoring.

Yes, multi-language code analysis supports Python, TypeScript, JavaScript, Go, and Rust. It uses language-aware parsers to produce structural summaries suitable for bug tracing and code audits across these languages.

The best way to generate a call graph and AST for a large project is using depth modes like overview, deep, or full. These modes balance structural detail against your token budget during codebase analysis.

Yes, you need file-system access to generate a control flow graph. Language-aware parsers require direct file-system access to read source files and produce accurate CFG and data flow graph representations.

To start a deep code analysis, point the tool to your repository path and select the deep depth mode. This instructs the parser to produce detailed AST and program dependence graphs within your token budget.

Reading raw source files obscures structural relationships because it consumes too many tokens and lacks explicit connections. Generating a program dependence graph and call graph instead clarifies these dependencies efficiently.