LibreAI / RAG et connaissances / code-graph-rag

// rag · vitali87/code-graph-rag

code-graph-rag

code-graph-rag est un projet de RAG et connaissances open source, créé par vitali87 (vitali87/code-graph-rag sur GitHub). Il compte 5 238 étoiles sur GitHub et 699 forks, il est publié sous la licence MIT et a été mis à jour pour la dernière fois sur GitHub le 8 octobre 2026.

RAG for your monorepo: query, understand and edit multi-language codebases with a knowledge graph.

★ 5,2 k étoiles⑂ 699 forksMITmis à jour oct. 2026

Voir sur GitHub ↗Official site ↗

▦ visuel — le README de ce projet ne contient aucune capture intégrable,
voici donc un emplacement réservé. Le guide ci-dessous est complet malgré tout.

// détails du projet

Détails du projet

Dépôt: vitali87/code-graph-rag
Étoiles GitHub: 5 238
Forks: 699
Licence: MIT
Dernière mise à jour sur GitHub: 8 octobre 2026
Catégorie: RAG et connaissances
Type de fiche: Guide d'installation complet
Site officiel: Official site

// fonctionnement

Comment ça marche

Clarify your codebase problem
Decide what you need: understanding a monorepo, finding dead code, tracing behavior, or safe automated edits.
Install the tool
Install code-graph-rag with the Tree-sitter and semantic extras via uv or pipx.
Start the stack and index
Start the packaged daemon, then parse your repository into the graph.
Ask your first question
Query the codebase in natural language and inspect the grounded, graph-backed answer.

// fonctionnalités

Pourquoi on l'adore

Knowledge-graph code index

Parses source with Tree-sitter into AST-level graph structure — functions, classes, calls and references — stored in Memgraph.

Natural-language querying

Asks questions in plain language; an AI model generates Cypher against the graph and answers grounded in the real structure.

Surgical code editing

Edit code through the agent with AST-based patching and a diff preview before anything changes.

Dead-code and structural analysis

Finds dead code by walking call and reference edges from entry points, and searches/rewrites by AST pattern with ast-grep.

Runtime overlay

cgr trace merges actually-executed calls (test runs or production eBPF profiles) into the graph, exposing dynamic dispatch static analysis cannot see.

MCP server mode

Exposes the graph to MCP-compatible agents, with incremental indexing that keeps the graph current as code changes.

// lancez-le

Lancez-le vous-même

Clone the repository
Clone Code-Graph-RAG if you want the source; most users install the tool.
```
git clone https://github.com/vitali87/code-graph-rag.git
cd code-graph-rag
```
Install with extras
Install the packaged tool with full Tree-sitter language support and semantic search.
```
uv tool install "code-graph-rag[treesitter-full,semantic]"
```
Index and query
Start the packaged Memgraph + Qdrant stack, parse your repo into the graph, and start querying.
```
cgr daemon up
cgr start --repo-path /path/to/repo --update-graph
```

// cas d'usage

Ce que vous pouvez créer

Understanding how a large monorepo fits together during onboarding
Finding provably dead code before a cleanup sprint
Answering 'what calls this function across all services?' in one query
Letting an agent make reviewed, AST-level edits across a multi-language codebase

// faq

Questions fréquentes

Who maintains Code-Graph-RAG?

It is maintained by vitali87 and contributors, with enterprise services offered around the open-source core.

Is it free?

Yes — the tool is MIT licensed and runs locally; enterprise cloud and on-premise services are optional paid offerings.

Does it need a paid API key?

Querying uses an AI model to generate Cypher, so you need access to a model provider you configure; the indexing itself is local.

How does it compare to Semble or grep-based search?

Search tools return chunks; Code-Graph-RAG builds a structural knowledge graph you can traverse — call edges, references, runtime traces — and edit through, at the cost of a heavier indexing stack.

How do I get started?

Les autres projets de la catégorie RAG et connaissances avec le plus d'étoiles du catalogue LibreAI — la sélection réaliste d'alternatives si code-graph-rag ne vous convient pas.

MarkItDown★ 193,5 k
MarkItDown — Microsoft tool that converts PDFs, Office files, images and more to clean Markdown for LLMs, MIT
Dify★ 157,8 k
Build Agentic workflows, RAG pipelines, with rich AI model and tool support on one collaborative workspace. Deploy on cloud, VPC, or…
ragflow★ 91,6 k
RAGFlow is a leading open-source Retrieval-Augmented Generation (RAG) engine that fuses cutting-edge RAG with Agent capabilities to create a superior context…
PaddleOCR★ 90,8 k
Lightweight OCR toolkit turning PDFs and images into structured, LLM-ready data in 100+ languages
MinerU★ 82,5 k
MinerU — turns PDFs and Office documents into LLM-ready Markdown and JSON for agents and RAG pipelines, Apache-2.0 plus terms

// plus en rag

Plus en RAG et connaissances

Projets de la catégorie RAG et connaissances proches dans le catalogue, les plus proches en étoiles de code-graph-rag — continuez à explorer de part et d'autre.

llm-graph-builder★ 5,3 k
Build Neo4j knowledge graphs from unstructured data using LLMs — then chat with your graph.
AutoRAG★ 5,1 k
An agent that finds anything on your computer — federated search over your files, in place.
UltraRAG★ 5,7 k
A low-code MCP framework for building complex, innovative RAG pipelines — visibly, not as a black box.
OpenKB★ 4,8 k
OpenKB compiles your documents into a persistent, interlinked wiki knowledge base — no vector DB
helix-db★ 6,1 k
An OLTP graph database with native vector and full-text search, built in Rust on object storage.
infinity★ 4,7 k
AI-native database for hybrid search across dense, sparse, and tensor vectors plus full text

Obtenir code-graph-rag

Gratuit et ouvert : récupérez le code source, mettez une étoile au dépôt et commencez à créer.

Voir sur GitHub ↗← Plus d'outils RAG et connaissances