← Back to Docs
Core Concepts
Understanding how SourcePrep processes and serves code context.
Local Indexing
How the semantic search engine works. Embeddings, chunking strategies, and the real-time file watcher.
Code Graph
The Rust-powered graph engine. How we map imports, function calls, and symbol definitions across your entire monorepo.
Context Assembly
How SourcePrep combines search results, trace neighbors, and path weights into a single optimized prompt payload for LLMs.
Graph Enrichment
ProA multi-step pipeline that keeps deepening your Code Graph — starting with fast structural parsing, then layering in meaning, module relationships, and architectural summaries. The graph tracks what it knows and flags when that knowledge goes stale, so your AI never acts on outdated context.