Super Agent

Super Agent is a powerful intelligent agent system built on the LangCrew framework, integrating browser automation, web search, code execution, file operations, and various other tools. It provides complete Web API services with streaming response support.

Prerequisites and Configuration

Environment Variables

Before using Super Agent, you need to configure environment variables.

Important Note: Please refer to the environment variable configuration example file in the project:

Configuration Example File: examples/components/web/super_agent/.env.example

Please configure all the following environment variables to ensure Super Agent runs properly:

# LLM API Key
OPENAI_API_KEY=

# E2B Sandbox Environment Configuration
E2B_API_KEY=
E2B_TEMPLATE=e2b-20250806-v1
E2B_DOMAIN=your-e2b-domain.example.com
E2B_TIMEOUT=3600

# S3 Storage Configuration
S3_ENDPOINT=
S3_AK=
S3_SK=
S3_BUCKET=
S3_GATEWAY=
S3_REGION=

# WebSearchTool Configuration

# WebSearchTool Configuration (required for web search functionality)
SERPER_API_KEY=your_serper_api_key_here

# ImageGenerationTool Configuration (optional, for AI image generation)
ARK_API_KEY=your_image_gen_api_key_here

Variable Descriptions

LLM API Key: OpenAI API key for the agent’s core reasoning capabilities
E2B Sandbox Environment Configuration: Used for secure code execution environment
- E2B_API_KEY: Requires applying for E2B API key
- E2B_TEMPLATE: To build a custom sandbox, refer to the official E2B documentation at https://e2b.dev/docs/sandbox-template. We provide a pre-configured template in langcrew/libs/langcrew-tools/e2b-template. After building your own, replace the E2B_TEMPLATE variable with your template ID.
- E2B_DOMAIN: Your E2B domain (e.g., your-e2b-domain.example.com)
- E2B_TIMEOUT: Sandbox timeout, set to 3600 seconds
S3 Storage Configuration: For file storage and image saving, requires configuring S3-compatible object storage service
Search Tool Configuration: For web search functionality, requires configuring search service endpoint and key
Image Generation Configuration: For AI image generation functionality, requires configuring image generation service key

Important Note: All environment variables should be configured according to your own service accounts and requirements. The template file .env.example provides examples of required variables.

Quick Start

Step 1: Navigate to the Example Project Directory

cd examples/components/web/super_agent

Step 2: Configure Environment Variables

Create a .env file in the current directory:

# Copy configuration example file
cp .env.example .env
# Edit the .env file to fill in necessary configurations

Step 3: Start Services

docker-compose -f compose-super-agent.yaml up --build

Step 4: Access Services

After startup is complete, you can access:

Super Agent API: http://localhost:8000
Frontend Web UI: http://localhost:3600

Note: The front-end service requires 5-6 minutes to start up.

Docker Compose Architecture Overview

The compose-super-agent.yaml file in the example directory defines the complete full-stack architecture:

frontend service:

Based on Node.js 18 Alpine image
Port mapping: 3600:3600
Automatically installs dependencies and starts development server
Environment variable: AGENT_API_HOST=http://super-agent:8000
Mounts ./web directory as frontend code

super-agent service:

Built based on the Dockerfile in the current directory
Port mapping: 8000:8000
Loads environment variables from .env file
Run mode: full, log level: info
Network: langcrew-network (for communication with frontend service)

Key Features

API Interfaces

Super Agent provides the following core API interfaces:

1. Chat Interface

POST /api/v1/chat
Content-Type: application/json

{
  "message": "Your task description",
  "session_id": "Session ID (optional, automatically generated if empty)"
}

Features:

Multi-turn conversation support
Real-time streaming response (Server-Sent Events)
Automatic session management
Tool invocation and execution monitoring

2. Stop Execution Interface

POST /api/v1/chat/stop
Content-Type: application/json

{
  "session_id": "Session ID to stop"
}

3. Update Task Interface

POST /api/v1/update_task
Content-Type: application/json

{
  "message": "New task instruction",
  "session_id": "Session ID"
}

Agent Capabilities

Browser Automation

Intelligent web page operation and data extraction based on browser-use
Automatic handling of complex multi-step web tasks and workflows
Intelligent page element recognition and interactive operations
Support for user takeover control and manual intervention for login pages
Built-in error recovery and retry mechanisms to ensure stable task execution

Information Retrieval and Processing

Web search and content scraping
Multi-source information integration and verification
Structured data extraction and analysis
Intelligent content summarization and report generation

Code Execution and File Operations

Python code execution in sandbox environment
Mathematical calculations, data analysis, chart generation
Complete file system operation support
Support for reading and writing various file formats

Multimedia Generation

AI-driven image generation
Support for various artistic styles and descriptions
Automatic image file saving and management

System Integration

Linux command execution support
Environment variable and configuration management
Multi-tool coordination and workflow orchestration

Usage Examples

Web Search and Information Extraction

curl -X POST "http://localhost:8000/api/v1/chat" \
  -H "Content-Type: application/json" \
  -d '{
    "message": "Search for 2024 artificial intelligence development trends and summarize key insights",
    "session_id": "a1b2c3d4e5f6g7h8"
  }'

Browser Automation Tasks

curl -X POST "http://localhost:8000/api/v1/chat" \
  -H "Content-Type: application/json" \
  -d '{
    "message": "Open the 12306 official website and help me check which high-speed trains from Beijing to Shanghai have available tickets after 3 PM the day after tomorrow",
    "session_id": "b2c3d4e5f6g7h8i9"
  }'

File Processing and Content Generation

curl -X POST "http://localhost:8000/api/v1/chat" \
  -H "Content-Type: application/json" \
  -d '{
    "message": "Create a research report on AI technology development and save it as a markdown file",
    "session_id": "c3d4e5f6g7h8i9j0"
  }'

Project Architecture

src/super_agent/
├── agent/
│   └── crew.py              # Main SuperAgentCrew class
├── config/
│   ├── config.py           # Configuration class definitions
│   └── prompt.py           # System prompts
├── main.py                 # Direct execution entry point
└── server.py              # FastAPI server

Integration with LangCrew Framework

Super Agent is built entirely on the LangCrew framework, showcasing the framework’s powerful capabilities:

Decorator Development Pattern: Uses @agent, @crew decorators
Tool Ecosystem: Integrates LangCrew-Tools toolset
Streaming Monitoring: Real-time execution status and result streams
Session Management: Persistent context and state

Super Agent is a perfect demonstration of the LangCrew framework’s capabilities, providing a complete reference implementation for building production-grade intelligent agent systems.