Awesome AI Agent Stack

RAG & Retrieval

Vector databases, search engines, embeddings and the frameworks that ground answers in your own data.

Vector databases

• chroma-core/chroma - Embedded, zero-config vector store; the right default for prototypes.
• qdrant/qdrant - Rust vector database with filtering and hybrid search; production-ready.
• pgvector/pgvector - Vector similarity search inside Postgres; skip a new database if you already run Postgres.
• lancedb/lancedb - Serverless, file-based vector database on Lance columnar format.
• milvus-io/milvus - Distributed vector database for billion-scale workloads.
• weaviate/weaviate - Vector database with built-in vectorizers and hybrid search.
• alibaba/zvec - A lightweight, lightning-fast, in-process vector database.
• VexDB-THU/VexDB-Lite - A cross-platform vector database integrable into existing applications.
• vearch/vearch - Distributed vector search for AI-native applications.
• vdaas/vald - Vald: a highly scalable distributed vector search engine.
• dingodb/dingo - A multi-modal vector database supporting upserts and vector queries.
• endee-io/endee - A high-performance vector database built for billions of vectors.
• philippgille/chromem-go - Embeddable vector database for Go with a Chroma-like interface.
• oceanbase/seekdb - The AI-native search database for agent storage, unifying vector search.
• Stevenic/vectra - Local vector database for Node.js, Pinecone-like but file-based.
• epsilla-cloud/vectordb - A high-performance vector database leveraging parallel graph computing.
• sdan/vlite - A fast vector database built in numpy.
• Semafind/semadb - No-fuss multi-index hybrid vector database and search engine.
• nuclia/nucliadb - NucliaDB, the AI search database for RAG.
• infiniflow/infinity - AI-native database for LLM applications with vector and full-text search.
• objectbox/objectbox-java - On-device database for Android and JVM with vector search.

Vector search libraries

• eminsk/nanovector - The SQLite of vector search and episodic memory for AI agents in ~120KB.
• spotify/annoy - Approximate nearest neighbors for static vectors.
• facebookresearch/faiss - A library for efficient similarity search and clustering of dense vectors.
• unum-cloud/USearch - Fast open-source search and clustering engine for vectors and more.
• nmslib/hnswlib - Header-only C++/Python library for fast approximate nearest neighbors.
• microsoft/DiskANN - Vector indexing library for fast, fresh, and filtered search at scale.
• sqliteai/sqlite-vector - SQLite-Vector: cross-platform, ultra-efficient SQLite extension.
• StarTrail-org/LEANN - Storage-efficient vector index for running RAG over personal data on a laptop.
• RyanCodrai/turbovec - Rust vector index with Python bindings built on TurboQuant compression.

Search engines

• quickwit-oss/tantivy - Full-text search engine library in Rust.
• elastic/elasticsearch - Distributed search and analytics engine.
• opensearch-project/OpenSearch - Search and analytics suite, Apache 2.0 fork.
• apache/solr - Enterprise full-text search platform.
• searxng/searxng - Free metasearch engine aggregating results without tracking or profiling.
• meilisearch/meilisearch - A lightning-fast search engine API bringing AI-powered hybrid search to your sites and applications.
• marqo-ai/marqo - Ecommerce Search and Discovery - marqo.ai.
• vespa-engine/vespa - Large-scale AI search platform for hybrid text and vector retrieval at internet scale.
• typesense/typesense - Fast typo-tolerant search engine with vector and hybrid search plus built-in RAG.
• manticoresoftware/manticoresearch - Open-source search database for full-text, vector, and hybrid search.
• oramasearch/orama - A complete search engine and RAG pipeline for browser and server.
• DeepBlueDynamics/lume - A small, crazy-fast hybrid search engine written in Rust.
• nixiesearch/nixiesearch - Hybrid search engine combining text and semantic search.
• cameodb/cameodb - CameoDB: the unified hybrid-search database engine.
• paradedb/paradedb - Postgres extension adding BM25 full-text search and hybrid retrieval.

Embeddings, rerankers & search

• huggingface/sentence-transformers - Standard library for embedding models and cross-encoder rerankers.
• FlagOpen/FlagEmbedding - BGE embeddings and rerankers; strong open baselines.
• AnswerDotAI/RAGatouille - ColBERT late-interaction retrieval made easy.
• xhluca/bm25s - Fast BM25 in Python for the lexical half of hybrid search.
• NovaSearch-Team/RAG-Retrieval - Unify efficient fine-tuning of RAG retrieval: embeddings and rerankers.
• castorini/rank_llm - RankLLM: Python toolkit for reproducible information retrieval research.
• ielab/llm-rankers - Document ranking with large language models: pointwise, listwise, and setwise.
• huggingface/text-embeddings-inference - A blazing fast inference solution for text embeddings models.
• castorini/pyserini - Pyserini: Python toolkit for reproducible IR research with sparse and dense retrieval.
• michaelfeil/infinity - Infinity: high-throughput, low-latency serving for embeddings and rerankers.
• superlinked/sie - Open-source inference server and production cluster for agent models.
• AnswerDotAI/rerankers - Unified API for reranker models in retrieval pipelines.
• dorianbrown/rank_bm25 - BM25 ranking for lexical search in Python.
• stanford-futuredata/ColBERT - Late-interaction neural retrieval for RAG.
• qdrant/fastembed - Fast, lightweight Python library for state-of-the-art text embeddings.

RAG frameworks & techniques

• neuml/txtai - All-in-one AI framework for semantic search and LLM orchestration.
• OSU-NLP-Group/HippoRAG - [NeurIPS'24] RAG framework inspired by human long-term memory for deeper multi-hop reasoning.
• carriex/recomp - RECOMP: compress and selectively augment retrieved documents for RAG.
• HKUDS/RAG-Anything - All-in-one multimodal RAG framework.
• NirDiamant/RAG_Techniques - Advanced RAG techniques, each with runnable code.
• Shubhamsaboo/all-rag-techniques - Every RAG technique, implemented simply.
• pguso/rag-from-scratch - Build RAG from scratch with local models, no black boxes.
• jamwithai/production-agentic-rag-course - Production agentic RAG course.
• upstash/context7 - Up-to-date library documentation served to LLMs and editors.
• pathwaycom/pathway - Python ETL framework for stream processing, real-time analytics, LLM pipelines, and RAG.
• pathwaycom/llm-app - Ready-to-run cloud templates for RAG, AI pipelines, and enterprise search with live data.
• The-Vibe-Company/quivr - Opiniated RAG for integrating GenAI in your apps 🧠 Focus on your product rather than the RAG.
• onyx-dot-app/onyx - Open Source AI Platform - AI Chat with advanced features that works with every LLM.
• infiniflow/ragflow - RAGFlow is a leading open-source Retrieval-Augmented Generation (RAG) engine that fuses.
• zylon-ai/private-gpt - Complete API layer for private AI applications on local models: RAG, skills, tools, MCP.
• VectifyAI/PageIndex - PageIndex: Document Index for Vectorless, Reasoning-based RAG.
• llmware-ai/llmware - Unified framework for enterprise RAG pipelines with small specialized models.
• OpenBMB/UltraRAG - Low-code MCP framework for building complex, innovative RAG pipelines.
• Azure/GPT-RAG - Enterprise-grade accelerator for agentic RAG on Azure.
• memfreeme/memfree - MemFree: hybrid AI search engine and AI page generator.
• ggozad/haiku.rag - Agentic RAG for local and self-hosted document search with hybrid retrieval.
• notadev-iamaura/OneRAG - Production-ready RAG framework with 1-line config swaps for providers.
• MoMoM101/RAG-ReActAgent - Retrieval-augmented generation system with a ReAct agent loop.
• flamehaven01/Flamehaven-Filesearch - Self-hosted RAG search engine: 34 formats, BM25 plus hybrid search.
• hanxiao/searchbox - Airgapped closed-corpus QA: a self-hosted agent explores documents and answers.
• nshkrdotcom/rag_ex - Elixir RAG library with multi-LLM routing across Gemini, Claude, and OpenAI.
• weizhepei/InstructRAG - [ICLR 2025] InstructRAG: instructing RAG via self-synthesized rationales.
• Tencent/WeKnora - LLM knowledge platform that turns documents into queryable RAG and agent workflows.
• MODSetter/SurfSense - Privacy-focused open source NotebookLM alternative with RAG over your own sources.
• zilliztech/deep-searcher - Deep research over private data using reasoning models and vector search.
• labring/FastGPT - FastGPT is a knowledge-based platform built on the LLMs, offers a comprehensive suite of.
• Minima-AI-Inc/minima - On-premises conversational RAG with configurable containers.
• netease-youdao/QAnything - Question answering over documents with a web UI.
• SciPhi-AI/R2R - Production-ready agentic RAG system with a dashboard.
• morphik-org/morphik-core - Open-source multimodal retrieval engine for documents, images and video.

Graph RAG

• HKUDS/LightRAG - Simple, fast graph-based RAG.
• microsoft/graphrag - Microsoft's graph-based RAG for global questions over corpora.
• OpenSPG/KAG - Logical-form-guided reasoning and retrieval framework over knowledge graphs.
• benmaster82/Kwipu - Ask questions across your Markdown notes using fully local Graph RAG.
• pingcap/autoflow - Graph RAG-based conversational knowledge base.
• NVIDIA/context-aware-rag - Context-aware RAG library for knowledge graph ingestion and retrieval.
• vitali87/code-graph-rag - The ultimate RAG for your monorepo: query and edit multi-language codebases.
• hhy-huang/HiRAG - [EMNLP'25] HiRAG: easy-to-use Graph RAG with hierarchical retrieval.
• raphaelmansuy/edgequake - High-performance GraphRAG inspired by LightRAG.
• 1517005260/graph-rag-agent - GraphRAG agent fusing GraphRAG, LightRAG, and Neo4j graph construction.
• automataIA/graphrag-rs - GraphRAG-rs: high-performance Rust implementation of GraphRAG.
• LHRLAB/HyperGraphRAG - [NeurIPS 2025] HyperGraphRAG: retrieval-augmented generation over hypergraphs.
• bibinprathap/VeritasGraph - VeritasGraph: open-source knowledge graph and GraphRAG framework.
• DerwenAI/strwythura - Strwythura: construct entity-resolved knowledge graphs from structured data.
• PorunC/CodeWiki - CodeWiki: knowledge platform analyzing repositories into AST graphs.
• Muvon/octocode - Structural code intelligence for AI agents: semantic search and knowledge graphs.
• gibram-io/gibram - GibRAM: in-memory knowledge graph server designed for retrieval.
• lyon-industries/graphrag-workbench - Interactive 3D visualization of knowledge graphs generated by GraphRAG.
• Abhishek-Aditya-bs/CodeGraph - GraphRAG over any codebase: a Neo4j knowledge graph of code entities.
• FareedKhan-dev/agentic-knowledge-graph - Knowledge-graph RAG built from structured metadata instead of LLM extraction.
• hanjiale/Temporal-GraphRAG - Official code for 'RAG Meets Temporal Graphs': time-sensitive modeling.
• yonk-labs/pg-raggraph - PostgreSQL-native GraphRAG: vector search, full-text, and knowledge graphs.
• awslabs/unified-kg-rag-on-aws - AWS-native knowledge graph RAG framework unifying graph-retrieval approaches.
• machuangtao/LLM-KG4QA - [EMNLP2025] LLM-KG4QA: large language models and knowledge graphs for QA.
• AmirhosseinHonardoust/Graph-RAG-Engine - An explainable AI system combining graph intelligence and vector search.
• Aliu-AiRobot/ESEILANE - High-performance knowledge graph engine for AI, LLMs, and GraphRAG.
• intuit/infigraph - AST-powered code intelligence engine: graph database plus hybrid semantic search.
• neo4j-labs/llm-graph-builder - Builds Neo4j knowledge graphs from unstructured documents using LLMs.

Semantic caching

• zilliztech/GPTCache - Semantic cache for LLMs, fully integrated with LangChain and LlamaIndex.
• aqstack/mimir - A drop-in proxy caching LLM API responses using semantic similarity.
• messkan/prompt-cache - Cut LLM costs by up to 80% with sub-millisecond cached responses.
• sensoris/semcache - Semantic caching layer for your LLM applications.
• vcache-project/vCache - Reliable and efficient semantic prompt caching with vCache.
• botirkhaltaev/semanticcache - A Go library for semantic caching with LRU eviction.
• nickleodoen/ferrocache - A distributed semantic cache service for LLM applications.
• zakariaf/RAG-Cache - High-performance LLM query cache with semantic search.
• Cozymori/VectorWave - One decorator for semantic caching plus golden-data regression testing.
• mpaymenremora/CacheCraft - Semantic cache strategy bench: replay recorded request traces.

RAG evaluation & benchmarks

• zilliztech/VectorDBBench - Benchmark suite for vector databases.
• hseb-benchmark/hseb - HSEB: benchmark for hybrid search engines.
• xmpuspus/kb-arena - Benchmark 19 retrieval architectures: vector, contextual, and knowledge-based.
• agentset-ai/reranker-eval - Benchmarking rerankers in a RAG pipeline on accuracy metrics.
• RUC-NLPIR/FlashRAG - Python toolkit for reproducing and benchmarking efficient RAG pipelines.