graphrag-rs

Why we included this project

This Rust implementation of GraphRAG turns unstructured documents into a persistent knowledge graph you can query in plain language. Answers are built from the named entities and the relations between them rather than pure vector similarity, so questions about connections between facts feel more grounded. The project ships with a zero-config CLI; `graphrag index` ingests a document and `graphrag ask` answers from the graph, with an `--ollama` flag to switch in local LLM quality entity extraction. You can also raise the bar with GLiNER or custom chunking when the defaults fall short. The more distinctive part is deployment: server-only, a fully client-side WASM build that runs entirely in the browser, or a hybrid of the two, plus optional WebGPU acceleration. That makes it a genuinely flexible base if you want a fast, self-contained RAG stack without a heavyweight cloud dependency.