The Complete Guide to Building AI Agent Systems in Go
AgenticGoKit is a production-ready Go framework for building intelligent agent workflows with dynamic tool integration, multi-provider LLM support, and enterprise-grade patterns.
New to AgenticGoKit? Start here:
- Build Your First Agent - Working agent in 5 minutes
- Core Concepts - Understand the fundamentals
- Examples - Real-world code patterns
v1beta is the modern, production-ready API for AgenticGoKit. All new projects should use v1beta.
- Getting Started Guide - Install and build your first agent
- Core Concepts - Agents, workflows, memory, tools
- Configuration - Environment setup and options
- Quick Reference - Common patterns and snippets
Working code for common patterns:
- Basic Agent - Simple chat agent
- Streaming Agent - Real-time streaming responses
- Sequential Workflow - Step-by-step pipeline
- Parallel Workflow - Concurrent execution
- DAG Workflow - Complex dependencies
- Loop Workflow - Iterative refinement
- Subworkflows - Nested workflows
- Memory & RAG - Knowledge bases
- Custom Handlers - Custom business logic
- Memory & RAG - Memory providers, vector search, knowledge bases
- Tool Integration - Adding tools to agents
- Custom Handlers - Implement custom logic
- Error Handling - Robust error patterns
- Performance - Optimization and tuning
- Troubleshooting - Common issues and solutions
- Migration Guide - Migrate from legacy APIs to v1beta
- API Versioning - Version strategy and stability
Perfect for developers new to AgenticGoKit:
- Build Your First Agent - 5-minute start
- Basic Agent Example - Complete working code
- Core Concepts - Understanding the framework
- Configuration Guide - Setup and options
Build production-ready systems:
- Streaming Agent - Real-time responses
- Sequential Workflow - Multi-step pipelines
- Memory & RAG - Add knowledge capabilities
- Tool Integration - Connect external tools
Master complex patterns:
- Parallel Workflows - Concurrent execution
- DAG Workflows - Complex dependencies
- Subworkflows - Nested patterns
- Performance Tuning - Optimization strategies
The following documentation covers legacy APIs and the agentcli tool. For new projects, use v1beta documentation instead.
Click to expand legacy documentation
- 5-Minute Quickstart - Get running immediately
- Your First Agent - Build a simple agent from scratch
- Multi-Agent Collaboration - Agents working together
- Memory & RAG - Add knowledge capabilities
- Tool Integration - Connect external tools
- Production Deployment - Deploy to production
- Agent Fundamentals - Understanding AgentHandler interface and patterns
- Memory & RAG - Persistent memory, vector search, and knowledge bases
- Multi-Agent Orchestration - Orchestration patterns and API reference
- Orchestration Configuration - Complete guide to configuration-based orchestration
- Examples & Tutorials - Practical examples and code samples
- Tool Integration - MCP protocol and dynamic tool discovery
- LLM Providers - Azure, OpenAI, Ollama, and custom providers
- Configuration - Managing agentflow.toml and environment setup
- Advanced Patterns - Advanced orchestration patterns and configuration
- RAG Configuration - Retrieval-Augmented Generation setup and tuning
- Memory Provider Setup - PostgreSQL, Weaviate, and in-memory setup guides
- Workflow Visualization - Generate and customize Mermaid diagrams
- Production Deployment - Scaling, monitoring, and best practices
- Error Handling - Resilient agent workflows
- Custom Tools - Building your own MCP servers
- Performance Tuning - Optimization and benchmarking
- Core Package API - Complete public API reference
- Agent Interface - AgentHandler and related types
- Memory API - Memory system and RAG APIs
- MCP Integration - Tool discovery and execution APIs
- CLI Commands - agentcli reference
Build production systems for:
- Customer Support: Multi-agent systems with memory and tool integration
- Content Generation: Sequential and parallel workflows for content creation
- Research Systems: RAG-powered agents with knowledge base access
- Data Processing: ETL pipelines with validation and transformation
- Code Analysis: Iterative refinement loops for quality assurance
Built from the ground up in Go for maximum performance and reliability:
- Goroutine-based concurrency for parallel workflows
- Minimal memory footprint and fast startup times
- Native compilation for deployment anywhere
Enterprise-grade features out of the box:
- Comprehensive error handling and recovery
- Structured logging and observability hooks
- Memory management and resource cleanup
- Type-safe APIs with full IDE support
Work with any LLM provider:
- OpenAI (GPT-4, GPT-3.5)
- Azure OpenAI
- Ollama (local models)
- Anthropic Claude
- Custom providers via simple interface
Build exactly what you need:
- Single agents or complex multi-agent systems
- Sequential, parallel, or DAG-based workflows
- Custom handlers for business logic
- Pluggable memory and tool providers
go get github.com/agenticgokit/agenticgokit/v1betaRequirements:
- Go 1.21 or higher
- LLM provider API key (OpenAI, Azure, etc.) or local Ollama installation
Want to contribute to AgenticGoKit? See our Contributor Documentation:
- Contributor Guide - Development setup and workflow
- Code Style - Coding standards and conventions
- Testing - Testing strategies and guidelines
- Adding Features - Feature development process
- Documentation Standards - Writing great docs
# Quick start for contributors
git clone https://github.com/agenticgokit/agenticgokit.git
cd agenticgokit
go mod tidy
go test ./...- GitHub Discussions - Ask questions and share ideas
- Issues - Report bugs and request features
- Contributing - How to contribute
- Changelog - Release notes and version history
- Roadmap - Planned features and improvements
- Design Documents - Architecture and design decisions