The ultimate LLM/AI application development framework in Go.

AI Prompts & Endpoints
Agent Skills View CodeWiki Knowledge Base

Eino

Eino is a Go-based LLM application development framework by ByteDance.

It provides component abstractions, a graph/chain orchestration engine,

streaming primitives, a callback system, and an Agent Development Kit (ADK)

for building production-grade LLM applications.

Repositories

- eino — core framework (this repo)
- eino-ext — component integrations (OpenAI, Ark, Ollama, Redis, S3, …)
- eino-examples — runnable example applications

Overview & Background

- Overview
- ByteDance Eino Practice
- Eino Open Source
- Graph or Agent — when to use which

Quick Start

- Simple LLM Application
- Agent with Tools
- Eino Cookbook

Core Concepts — Components

Components are the typed building blocks of eino pipelines. Each has a defined
interface in the core repo; implementations live in eino-ext.

- Components overview
- ChatModel
- ChatTemplate
- ToolsNode
- How to create a Tool
- Retriever
- Indexer
- Embedding
- DocumentLoader
- DocumentParser
- DocumentTransformer
- Lambda
- AgenticChatModel
- AgenticChatTemplate
- AgenticToolsNode

Core Concepts — Orchestration

The orchestration layer composes components into executable pipelines.
Chain is a linear sequence; Graph is a DAG with conditional edges;
Workflow is a higher-level structured abstraction over Graph.

- Chain & Graph introduction
- Orchestration design principles
- Workflow orchestration framework
- Stream programming essentials
- Callback system
- CallOption capabilities
- Checkpoint & interrupt/resume

Core Concepts — Flow Integration

- ReAct agent
- Multi-agent hosting

Agent Development Kit (ADK)

The ADK provides a higher-level runtime for building, composing, and deploying
agents. It sits above the Graph layer and introduces Agent, Skill, and
Middleware abstractions.

- ADK overview
- Agent quickstart
- Agent interface
- Agent collaboration
- Agent implementations
- ChatModel agent
- Workflow agent
- Supervisor agent
- Plan-and-execute agent
- Deep agents
- Agent extension
- Human-in-the-loop (HITL)
- ADK callbacks
- ChatModelAgent middleware
- Filesystem middleware
- Skill middleware
- Summarization middleware
- Plan-task middleware
- Tool-search middleware
- Tool-reduction middleware
- Patch-toolcalls middleware

DevOps Tooling

- IDE plugin guide
- Visual orchestration plugin
- Visual debug plugin

Ecosystem Integrations (eino-ext)

- ChatModel integrations
- Document integrations
- Embedding integrations
- Tool integrations
- Callback integrations
- Indexer integrations
- Retriever integrations
- ChatTemplate integrations

Release Notes & Migration

- v0.1
- v0.2
- v0.3 — breaking changes
- v0.4 — compose optimization
- v0.5 — ADK implementation
- v0.6 — JSON schema optimization
- v0.7 — interrupt/resume refactor
- v0.8 — ADK middlewares

FAQ

- FAQ