gorag

🦖 GoRAG

GoRAG is a production-ready, high-performance RAG (Retrieval-Augmented Generation) framework built entirely in Go. Designed for enterprise scalability, it seamlessly connects your internal data to the most powerful LLMs with zero Python dependencies.

English | 中文文档

🔥 Why GoRAG?

Stop fighting with Python dependency hell and slow async loops. GoRAG brings the power of Go's concurrency and static typing to the AI world.

• 🚀 Blazing Fast: Process 100M+ files with 10 built-in concurrent workers.
• 🛡️ Enterprise Ready: Built-in Circuit Breakers, Graceful Degradation, Observability, and Metrics.
• 🧩 Highly Modular: Swap LLMs, Vector Stores, and Parsers with a single line of code.
• 🧠 Advanced Retrieval: Supports Multi-hop RAG, Agentic RAG, Semantic Chunking, HyDE, and RAG-Fusion.
• ☁️ Cloud Native: Compiles to a single binary. Perfect for Kubernetes deployments.
• 📦 No External DB Required: Ships with native pure-Go vector database govector for zero-setup local deployments, while supporting enterprise databases like Milvus, Qdrant, and Pinecone.

✨ Latest Updates (v1.0.0)

• Complete LLM SDK Migration: Fully powered by the unified gochat SDK. Write once, seamlessly switch between OpenAI, Anthropic, Ollama, Azure, and domestic Chinese LLMs.
• Native Vector Database Integration: Integrated govector natively for a pure Go, zero-dependency embedded vector search experience.
• Extensive Parser Ecosystem: Now supports 16 document types natively out-of-the-box (Text, PDF, DOCX, HTML, Email, Code, etc.) and offers independent plugins for Audio, Video, and Webpages.

🛠️ Out-of-the-Box Support

🤖 Supported LLMs (Powered by gochat)

• OpenAI: GPT-4o, GPT-4 Turbo, GPT-3.5
• Anthropic: Claude 3 (Opus, Sonnet, Haiku)
• Local/Open Source: Ollama (Llama 3, Qwen, Mistral)
• Enterprise: Azure OpenAI
• Compatible APIs: Kimi, DeepSeek, GLM-4, Minimax, Baichuan, and more.

🗄️ Supported Vector Stores

• govector 🌟 (Native pure-Go embedded vector DB - Zero setup!)
• Milvus (Enterprise standard)
• Qdrant (High performance)
• Weaviate (Semantic search engine)
• Pinecone (Fully managed cloud DB)
• Memory (For quick testing/dev)

🚀 Quick Start

Installation

go get github.com/DotNetAge/gorag

1. The 10-Line RAG Setup

Get a complete RAG pipeline running locally using Ollama and our native pure-Go govector database. No external databases or API keys required!

package main

import (
	"context"
	"log"

	"github.com/DotNetAge/gochat/pkg/client/base"
	"github.com/DotNetAge/gochat/pkg/client/ollama"
	"github.com/DotNetAge/gorag/rag"
	"github.com/DotNetAge/gorag/vectorstore/govector"
)

func main() {
	ctx := context.Background()

	// 1. Initialize LLM Client (Powered by gochat)
	llmClient, _ := ollama.New(ollama.Config{
		Config: base.Config{Model: "qwen:0.5b"},
	})

	// 2. Initialize Native Go Vector Store (Zero setup)
	vectorStore, _ := govector.NewStore(govector.Config{
		Dimension:  1536,
		Collection: "my_knowledge",
	})

	// 3. Create RAG Engine
	engine, _ := rag.New(
		rag.WithLLM(llmClient),
		rag.WithVectorStore(vectorStore),
	)

	// 4. Index a Document
	engine.Index(ctx, rag.Source{
		Type:    "text",
		Content: "GoRAG is a native Go framework for Retrieval-Augmented Generation.",
	})

	// 5. Query
	resp, _ := engine.Query(ctx, "What is GoRAG?", rag.QueryOptions{TopK: 3})
	log.Println("Answer:", resp.Answer)
}

2. High-Performance Directory Indexing

Need to ingest an entire codebase or documentation folder? GoRAG handles it automatically with 10 concurrent workers.

// ... (engine initialization)

// 🚀 Index an entire directory! Automatically detects file types (.pdf, .go, .md, .docx)
err := engine.IndexDirectory(ctx, "./my-company-docs")
if err != nil {
    log.Fatal(err)
}

// Stream the answer for better UX
ch, _ := engine.QueryStream(ctx, "Summarize the Q3 financial reports", rag.QueryOptions{
    Stream: true,
})

for resp := range ch {
    fmt.Print(resp.Chunk)
}

📊 Performance Benchmarks

GoRAG leaves Python-based frameworks in the dust. Tested on a standard Intel Core i5 (No GPU):

Note: Enabling GPU acceleration via Milvus/Ollama provides an additional 3x-5x performance boost.

🌟 Advanced RAG Patterns

GoRAG isn't just a wrapper; it implements state-of-the-art AI retrieval techniques natively:

• Agentic RAG: Let the LLM autonomously decide which tools to use and when to retrieve data.
• Multi-hop RAG: Breaks down complex questions requiring multi-step reasoning.
• Semantic Chunking & HyDE: Intelligent document splitting and Hypothetical Document Embeddings for superior recall.

🛠 CLI Usage

GoRAG ships with a powerful CLI out of the box:

# Install the CLI
go install github.com/DotNetAge/gorag/cmd/gorag@latest

# Index a document directly
gorag index --api-key $OPENAI_API_KEY --file ./docs/architecture.md

# Query your knowledge base
gorag query --api-key $OPENAI_API_KEY "How does the circuit breaker work?"

🤝 Contributing & Community

We are building the future of enterprise AI in Go. We welcome all contributions!

• Read our Contribution Guidelines
• Give us a ⭐️ if you find this project useful!

📄 License

MIT License. See LICENSE for more information.