MoFA Logo
M
O
F
A

让普通开发者成为全栈 AI 工程师

模块化 AI 代理框架

通过可组合的 AI 代理框架，让每个开发者都能轻松构建、调试和部署复杂的 AI 应用

5 min
Quick Setup
100+
Built-in Agents
∞
Combinations

Get Started in Seconds

One command to install MoFA and start building AI agents

Terminal
❯ ▊
# pip install mofa-stage # MoFA_Stage GUI
❯ ▊
# Clone an existing data flow
❯ ▊
# Start your data flow

Why Choose MoFA

Making AI development simple, efficient, and enjoyable

Composable Agent Architecture

Build complex AI applications by connecting agents via YAML-defined flows. Leverage a core kernel with modules for RAG (embedding, splitting, vector stores), planning, and tool integration. Easily orchestrate data flow between agents.

Rapid Agent Development

MoFA offers a clear structure for agent development, significantly reducing boilerplate and letting you focus on core logic. The MoFA Stage visual IDE further accelerates the entire development cycle, from creation to debugging. Get started in just 5 minutes.

Rich Agent Hub & Dev Tools

Access 100+ pre-built agents from our Agent Hub, covering data connectors, LLM integrations, and specialized tools. MoFA Stage further enhances development with visual agent management, an integrated terminal, and an advanced code editor.

Highly Extensible Framework

Easily write custom Python agents. Integrate third-party tools, models, and data sources through well-defined interfaces. Extend core functionalities like memory (e.g., Mem0 integration) or RAG strategies by implementing custom components.

Real Examples: AI Dataflows in Action

Explore different types of AI dataflows - from simple hello world to complex research automation

Hello World

Basic agent example

flowchart TB terminal-input[🖥️ Terminal Input<br/>User Query] agent[🤖 Agent<br/>Process & Respond] terminal-input --> agent agent --> terminal-input classDef inputNode fill:#e1f5fe,stroke:#0277bd,stroke-width:2px classDef agentNode fill:#f3e5f5,stroke:#7b1fa2,stroke-width:2px class terminal-input inputNode class agent agentNode
Code

ArXiv Research

Automated paper research

flowchart TB terminal[🖥️ Terminal Input<br/>Research Task] extractor[🔍 Keyword Extractor<br/>Extract Keywords] downloader[📥 Paper Downloader<br/>Download Papers] analyzer[🔬 Paper Analyzer<br/>Analyze Content] writer[✍️ Report Writer<br/>Generate Report] feedback[💬 Feedback Agent<br/>Review & Suggest] refinement[🔧 Refinement Agent<br/>Improve Report] terminal --> extractor extractor --> downloader downloader --> analyzer terminal --> analyzer analyzer --> writer terminal --> writer writer --> feedback terminal --> feedback feedback --> refinement terminal --> refinement classDef inputNode fill:#e8f5e8,stroke:#2e7d32,stroke-width:2px classDef processNode fill:#fff3e0,stroke:#f57c00,stroke-width:2px classDef analysisNode fill:#e3f2fd,stroke:#1976d2,stroke-width:2px classDef outputNode fill:#fce4ec,stroke:#c2185b,stroke-width:2px class terminal inputNode class extractor,downloader processNode class analyzer,feedback analysisNode class writer,refinement outputNode
Code

RAG System

RAG for Q&A

flowchart TB terminal[🖥️ Terminal Input<br/>User Question] retrieval[🔍 RAG Retrieval<br/>Search Knowledge] reasoner[🧠 Reasoner Agent<br/>Generate Answer] terminal --> retrieval retrieval --> reasoner terminal --> reasoner retrieval --> terminal reasoner --> terminal classDef inputNode fill:#e8f5e8,stroke:#2e7d32,stroke-width:2px classDef retrievalNode fill:#f3e5f5,stroke:#7b1fa2,stroke-width:2px classDef reasoningNode fill:#e1f5fe,stroke:#0277bd,stroke-width:2px class terminal inputNode class retrieval retrievalNode class reasoner reasoningNode
Code

GoSim Pedia

Multi-agent web research

flowchart TB openai[🤖 OpenAI Server<br/>Chat Interface] gosim[🎮 GoSim Pedia Agent<br/>Main Controller] firecrawl[🕷️ Firecrawl Agent<br/>Web Scraping] rag[🧠 GoSim RAG Agent<br/>Knowledge Retrieval] serper[🔍 Serper Search Agent<br/>Web Search] openai <--> gosim gosim --> firecrawl firecrawl --> gosim gosim --> rag rag --> gosim gosim --> serper serper --> gosim classDef serverNode fill:#e8f5e8,stroke:#2e7d32,stroke-width:2px classDef mainNode fill:#e1f5fe,stroke:#0277bd,stroke-width:2px classDef toolNode fill:#fff3e0,stroke:#f57c00,stroke-width:2px class openai serverNode class gosim mainNode class firecrawl,rag,serper toolNode
Code

Mem0 Memory System

Memory-enhanced dataflow

flowchart TB terminal[🖥️ Terminal Input<br/>User Task] retrieval[🧠 Memory Retrieval<br/>Fetch Context] reasoner[🤔 Reasoner<br/>Process & Think] record[💾 Memory Record<br/>Store Results] terminal --> retrieval retrieval --> reasoner terminal --> reasoner reasoner --> record terminal --> record retrieval --> terminal reasoner --> terminal record --> terminal classDef inputNode fill:#e8f5e8,stroke:#2e7d32,stroke-width:2px classDef memoryNode fill:#f3e5f5,stroke:#7b1fa2,stroke-width:2px classDef processNode fill:#e1f5fe,stroke:#0277bd,stroke-width:2px classDef storageNode fill:#fff3e0,stroke:#f57c00,stroke-width:2px class terminal inputNode class retrieval memoryNode class reasoner processNode class record storageNode
Code

Agent Creation System

Auto-generate AI agents

flowchart TB openai[🤖 OpenAI Server<br/>API Interface] config[⚙️ Config Generator<br/>Generate Settings] code[👨‍💻 Code Generator<br/>Write Agent Code] dependency[📦 Dependency Generator<br/>Manage Dependencies] openai --> config openai --> code config --> code openai --> dependency code --> dependency config --> dependency dependency --> openai classDef serverNode fill:#e8f5e8,stroke:#2e7d32,stroke-width:2px classDef generatorNode fill:#e1f5fe,stroke:#0277bd,stroke-width:2px classDef codeNode fill:#f3e5f5,stroke:#7b1fa2,stroke-width:2px classDef depNode fill:#fff3e0,stroke:#f57c00,stroke-width:2px class openai serverNode class config generatorNode class code codeNode class dependency depNode
Code

XiaoWang Multi-Agent

Multi-agent reflection & generation

flowchart TB terminal[🖥️ XiaoWang Terminal<br/>Task Input] dlc[🎯 Agent DLC<br/>Task Processing] generate[🔧 Agent Generate<br/>Content Creation] reflection[🤔 Agent Reflection<br/>Self-Improvement] terminal --> dlc dlc --> generate generate --> reflection reflection --> generate generate --> dlc dlc --> terminal generate --> terminal reflection --> terminal classDef inputNode fill:#e8f5e8,stroke:#2e7d32,stroke-width:2px classDef taskNode fill:#e1f5fe,stroke:#0277bd,stroke-width:2px classDef generateNode fill:#f3e5f5,stroke:#7b1fa2,stroke-width:2px classDef reflectNode fill:#fff3e0,stroke:#f57c00,stroke-width:2px class terminal inputNode class dlc taskNode class generate generateNode class reflection reflectNode
Code

← Scroll horizontally to explore more workflows →

See MoFA in Action

Watch how developers use MoFA to build sophisticated AI applications in minutes

Developer Hall of Fame

We celebrate everyone who made MoFA possible

  • MoFA Core Team
  • MoFA Contributors
  • Dora-rs Contributors
  • 2025 MoFA Hackathon Winners
  • 2024 MoFA Hackathon Winners
chengzi0103
XiaoKuge
betasecond
BH3GEI
alabulei1
SUNNYsyy2005
practicenotes007
tangdm9527
guofoo
cshbli
joulei
xforcevesa
starlitxiling
Gege-Wang
ZhangHanDong
Nely0
zerd1y
juntao
ROLFFFX
tdmlxq
jzzzzzzzzzzzzzzzz
451846939
heyong4725