🚀 DataAgent

中文 · English

Data + AI Agent: Enterprise Data Task Solution

🚀 DataAgent is a next-generation enterprise data intelligence platform for Data + AI scenarios, reimagining the entire data engineering pipeline through the Agent paradigm. Deeply integrating NL2SQL, unified semantic layers, and multi-agent collaboration, it delivers end-to-end data analysis and feature mining across financial risk control, AI for Science, and other core domains.

🌟 Why DataAgent

🏆 Scenario Advantages

⚡ Core Capabilities

📋 Environment Requirements

📚 Documentation

Full documentation lives under docs/ (中文 · English). Build and preview locally:

uv sync --extra mkdoc
uv run mkdocs serve -f docs/mkdocs.yml

🚴 Installation

1️⃣ Clone the project

git clone https://gitcode.com/datagallery/DataAgent.git
cd DataAgent

2️⃣ Install dependencies (uv recommended)

# Install dependencies
uv sync

# Activate virtual environment
source .venv/bin/activate  # Linux / macOS
.venv\Scripts\activate     # Windows

3️⃣ Or use pip

pip install -e .

4️⃣ Configure environment variables

# Copy environment template
cp .env.example .env

# Edit .env file with your actual configuration values

⚡ Quick Start

🎮 Interactive quick start

uv run -m dataagent quickstart

Follow the prompts to enter model configuration and start chatting with the Agent!

📁 Start with config file

# Terminal interactive mode
uv run -m dataagent --config dataagent/core/flex/examples/quickstart.yaml

🔍 Config check

# Check environment variable references in config
uv run -m dataagent config check dataagent/core/flex/examples/quickstart.yaml

📖 Usage

🐍 Python SDK

from dataagent import DataAgent

agent = DataAgent.from_config("path/to/config.yaml")

# Single-turn conversation
response = await agent.chat("Analyze sales data trends for the past week")
print(response)

# Streaming conversation
async for chunk in agent.astream(input={"user_query": "Generate user report"}):
    print(chunk, end="", flush=True)

📝 YAML Config Example

AGENT_CONFIG:
  name: "My Data Agent"
  version: "1.0"
  description: "Data Analysis Agent"
  backend: "langgraph"
  type: "react"

MODEL:
  chat_model:
    provider: "deepseek"
    model_type: "chat"
    params:
      model: "deepseek-chat"
      temperature: 0.7
      base_url: "$env{DEEPSEEK_BASE_URL}"
      api_key: "$env{DEEPSEEK_API_KEY}"

WORKSPACE:
  path: "/tmp/dataagent_workspace"
  allow_path:
    - "/tmp/dataagent_workspace"

🌐 A2A 1.0 Server Mode

# Start A2A server
uv run -m dataagent serve-a2a \
  --config path/to/config.yaml \
  --host 0.0.0.0 \
  --port 9999 \
  --auth-token your_token

# Service endpoints
# ├── 🌟 AgentCard: http://localhost:9999/.well-known/agent.json
# ├── 📡 JSON-RPC:  http://localhost:9999/a2a/jsonrpc
# └── 🔌 REST:      http://localhost:9999/a2a/rest

⚙️ Configuration

🔐 Environment Variables

📌 For more configuration, refer to .env.example

📄 License

This project is licensed under the Apache License 2.0 - see the LICENSE file for details.