Go-LLMs: Unified Go Library for LLM Integration

This was a good start and prototype to get the concept started quickly. This library has now been superceded by a rust based library available at https://github.com/lexlapax/rs-llmspell. Going forward, this library will not be actively worked on unless someone wants to add a quick fix if it's usefule to them.

Build powerful AI applications with a clean, unified interface to multiple LLM providers. Go-LLMs provides everything you need: from simple text generation to complex multi-agent workflows with built-in tools and structured outputs.

Why Go-LLMs?

• One Library, All Providers - Switch between OpenAI, Anthropic, Google, Ollama, and more with the same code
• Production Readiness Mindset - Built-in error handling, rate limiting, retries, and monitoring
• Rich Tooling - 33+ built-in tools for web, files, calculations, and data processing
• Smart Agents - Create conversational AI that can use tools and coordinate with other agents
• Structured Data - Get reliable, validated JSON/XML output instead of unpredictable text
• Go Native - Designed for Go developers with clean APIs and minimal dependencies

Features

✨ Unified Provider API - OpenAI, Anthropic, Google Gemini, Vertex AI, Ollama, OpenRouter
🛠️ Built-in Tool System - Web search, file operations, calculations, APIs, and more
🤖 Intelligent Agents - Conversational AI with memory, tools, and workflow orchestration
📊 Structured Outputs - JSON schema validation with automatic error recovery
🔍 Tool Discovery - Dynamic tool exploration perfect for scripting engines
🌐 Multimodal Content - Text, images, audio, video, and file support
⚡ Performance Optimized - Concurrent execution, streaming, caching

Installation

go get github.com/lexlapax/go-llms

Quick Start

Get started in 5 minutes with our interactive quickstart guide.

1. Simple AI Conversation

package main

import (
    "context"
    "fmt"
    "os"
    
    "github.com/lexlapax/go-llms/pkg/llm/provider"
    "github.com/lexlapax/go-llms/pkg/agent/core"
    "github.com/lexlapax/go-llms/pkg/llm/domain"
)

func main() {
    // Create provider
    p := provider.NewOpenAIProvider(os.Getenv("OPENAI_API_KEY"), "gpt-4")
    
    // Create agent
    agent := core.NewLLMAgent("assistant", "gpt-4", core.LLMDeps{Provider: p})
    agent.SetSystemPrompt("You are a helpful assistant.")
    
    // Chat
    state := domain.NewState()
    state.Set("user_input", "Explain quantum computing in simple terms")
    result, _ := agent.Run(context.Background(), state)
    
    fmt.Println(result.Get("response"))
}

2. Agent with Tools

// Create smart agent with built-in tools
agent := core.NewLLMAgent("smart-assistant", "gpt-4", core.LLMDeps{Provider: p})

// Add powerful tools
agent.AddTool(web.NewWebSearchTool(webAPIKey))
agent.AddTool(file.NewFileReadTool())
agent.AddTool(calculator.NewCalculatorTool())

// Agent can now search web, read files, and calculate
state := domain.NewState()
state.Set("user_input", "Search for recent Go releases and calculate days since Go 1.21")
result, _ := agent.Run(context.Background(), state)

3. Structured Data Extraction

// Define what you want
schema := &schema.Schema{
    Type: "object",
    Properties: map[string]*schema.Schema{
        "name":     {Type: "string"},
        "sentiment": {Type: "string", Enum: []interface{}{"positive", "negative", "neutral"}},
        "confidence": {Type: "number"},
    },
    Required: []string{"name", "sentiment"},
}

// Get reliable structured output
agent.SetSchema(schema)
state.Set("user_input", "Analyze this review: 'Amazing product, works perfectly!'")
result, _ := agent.Run(context.Background(), state)

// Guaranteed to match your schema
data := result.Get("structured_output")

4. Multi-Agent Workflows

// Create specialized agents
extractor := core.NewLLMAgent("extractor", "gpt-4", core.LLMDeps{Provider: p})
analyzer := core.NewLLMAgent("analyzer", "claude-3-sonnet-20240229", core.LLMDeps{Provider: claude})

// Orchestrate with workflows
workflow := workflow.NewSequentialAgent("data-pipeline", []domain.BaseAgent{
    extractor,  // First: extract data from text
    analyzer,   // Second: analyze extracted data
})

// Process data through the pipeline
state.Set("document", "Large document content...")
result, _ := workflow.Run(context.Background(), state)

Learning Resources

📖 Documentation

• Complete Documentation Hub - Start here for everything
• 5-Minute Quickstart - Get running immediately
• User Guide - Task-oriented guides with 5 learning paths
• Technical Documentation - Architecture and implementation details

🚀 Learning Paths

• Beginner Path - 5 simple projects to get started
• Developer Path - Build production applications
• Architect Path - Design robust systems
• Production Path - Deploy and scale

💡 Examples & Tutorials

• 80+ Working Examples - Provider, agent, tool, and workflow examples
• Built-in Tools Reference - Complete guide to 33+ built-in tools
• Tool Usage Examples - Practical patterns and integration examples
• Chat Applications Guide - Build conversational AI
• Data Extraction Guide - Reliable data processing
• Agent Communication - Multi-agent coordination

Supported Providers

OpenAI • Anthropic • Google Gemini • Google Vertex AI • Ollama • OpenRouter

Provider	Best For	Models	Setup
OpenAI	General use, reliability	GPT-4o, GPT-4 Turbo, GPT-4o-mini	Guide
Anthropic	Analysis, reasoning	Claude 3.5 Sonnet, Claude 3 Opus	Guide
Google Gemini	Multimodal, speed	Gemini 2.0 Flash Lite, Gemini Pro	Guide **
Vertex AI **	Enterprise, compliance	Gemini + partner models	Guide
Ollama	Local hosting, privacy	Llama, Mistral, CodeLlama	Guide
OpenRouter	Model variety, cost	400+ models (68 free)	Guide

** Untested integration

See our provider comparison guide for detailed feature matrices.

What's New

v0.3.6 (Latest) - Production-Ready Documentation 📚

Comprehensive documentation and quality assurance improvements:

• Complete Godoc Coverage - Enhanced 300+ Go files with comprehensive documentation
• Documentation Automation - Automated validation, link fixing, and content consistency
• Structural Reorganization - 95+ documentation files with proper navigation and learning paths
• Enhanced User Experience - Complete user guide with visual diagrams and task-oriented guides
• API Consistency - Updated 750+ code examples to current v0.3.5+ standards
• Technical Reference - Advanced tool patterns, integration guides, and architectural documentation

v0.3.5 - Complete Scripting Engine Support 🚀

Comprehensive bridge integration for go-llmspell and other scripting engines:

• Schema Repositories with versioning and persistence
• Enhanced Error Handling with serializable errors and recovery strategies
• Event System with serialization, filtering, and replay capabilities
• Tool Discovery with metadata-first exploration (33+ built-in tools)
• Workflow Serialization with templates and script-based execution
• Testing Infrastructure with mocks, fixtures, and comprehensive scenarios

Full release history in CHANGELOG.md.

Architecture

Go-LLMs uses a clean, modular architecture designed for reliability and extensibility:

pkg/
├── llm/         # Provider implementations and domain types
├── agent/       # Intelligent agents, tools, workflows, events
├── schema/      # JSON Schema validation and type conversion  
├── structured/  # Output parsing with error recovery
├── errors/      # Serializable error system with recovery
└── testutils/   # Comprehensive testing infrastructure

Design Principles:

• Unified Interfaces - Same API across all providers and components
• Fail-Safe Defaults - Graceful degradation and automatic error recovery
• Type Safety - Strong typing with schema validation throughout
• Performance First - Concurrent execution, streaming, and efficient state management
• Bridge Friendly - JSON-serializable types for scripting engine integration

Get Started

# Install
go get github.com/lexlapax/go-llms

# Try the quickstart
export OPENAI_API_KEY="your-key-here"
go run docs/user-guide/getting-started/quickstart.go

Choose your path:

• New to AI? → 5-Minute Quickstart
• Build apps? → Chat Application Guide
• Production use? → Enterprise Deployment
• Contributing? → Technical Documentation

Community & Support

• GitHub Issues - Bug reports and feature requests
• Discussions - Questions and community
• Contributing Guide - Development and contribution guidelines
• Documentation Style Guide - Standards for code documentation
• Changelog - Complete version history

Status

✅ Actively Maintained - Regular updates and improvements
✅ Comprehensive Testing - 280+ tests with >85% coverage
✅ Complete Documentation - User guides, API docs, examples
✅ Bridge Compatible - Ready for scripting engine integration

License: MIT - see LICENSE for details