Introduction
Carrot Agent is an intelligent agent framework developed in Go, designed for containerized operation with persistent memory and automatic skill learning capabilities.
🌟 Why Carrot Agent?
Unlike traditional chatbots, Carrot Agent:
- Remembers: Maintains context across sessions with hierarchical memory
- Learns: Automatically creates reusable skills from complex workflows
- Acts: Executes real-world tasks through tool calling
- Scales: Runs efficiently in containers with minimal resource usage
- Secures: Implements strict security constraints for tool operations
- Localizes: Supports both English and Chinese languages
Key Features
🧠 Hierarchical Memory System
Three-tier memory architecture ensures your agent never forgets important information:
- Snapshot Memory: Short-term contextual information
- Session Memory: Conversation history and state
- Long-term Memory: Persistent knowledge across sessions
🎯 Automatic Skill Learning
After completing complex tasks (5+ tool calls), the agent automatically:
- Analyzes the workflow
- Generates a reusable skill
- Saves it for future use
This means your agent gets smarter over time!
🔧 Powerful Tool Registry
Built-in tools include:
- File read/write operations (with path restrictions)
- HTTP requests (with URL validation)
- Memory management
- Skill CRUD operations
- System information
- Time utilities
All tools run with security constraints to prevent unauthorized access.
🐳 Container-First Design
- Official Docker images
- Non-root user execution
- Volume-based data persistence
- Health checks and auto-restart
- One-command deployment with Docker Compose
🤖 Multi-Model Support
Works seamlessly with:
- OpenAI GPT models (GPT-4, GPT-3.5)
- Anthropic Claude
- Any OpenAI-compatible API (OpenRouter, etc.)
🎨 Modern Web Interface
- Built with React, TypeScript, and Ant Design
- Responsive design for desktop and mobile
- Real-time chat interface with tool execution results
- Multi-language support (English and Chinese)
- Authentication for secure access
Use Cases
- Personal Assistant: Remember preferences and maintain context
- Development Helper: Code generation, file operations, documentation
- Research Agent: Web scraping, data collection, analysis
- Automation Workflows: Complex multi-step task automation
- Knowledge Management: Organize and retrieve information
Quick Example
bash
# Start with Docker Compose
docker-compose up -d
# Chat with your agent
curl -X POST http://localhost:8080/api/chat \
-H "Content-Type: application/json" \
-d '{
"message": "Help me create a Python script that reads a CSV file",
"session_id": "my-session"
}'The agent will:
- Understand your request
- Use appropriate tools (file operations, code generation)
- Save the workflow as a skill for future use
- Remember your preferences
Next Steps
- Quick Start - Get up and running in 5 minutes
- Installation - Detailed installation guide
- Architecture - Understand how it works
- API Reference - Explore the API endpoints