Skip to content
Docs

Code Indexing and AST Analysis Pipeline

Pipeline: Git Clone -> File Scan (streaming) -> AST Parse (concurrent tree-sitter) -> Graph Save (batched 1000) -> LLM Enrich (optional) -> Finalize.

Full AST: TypeScript/TSX, JavaScript/JSX, Python, Go, Java, Rust, Swift. Module-level (generic parser): Kotlin, C/C++, C#, Ruby, PHP, Scala, Vue, Svelte, COBOL.

Sources: Local path, Git URL (shallow clone, token auth), GitHub OAuth (Octokit), File upload (zip/folder, 512MB / 50K files / 25MB each).

Parsing: p-limit (default 4, max 32), 30s timeout, noise filtering, cross-file call graph, deterministic IDs, batch persistence.