AEnv Python SDK

A production-grade Python SDK for managing AI agent tools in containerized environments with MCP protocol support.

Features

• Simple Tool Definition: Create tools with just a decorator
• MCP Protocol Support: Full compatibility with Model Context Protocol using FastMCP
• Containerized Environments: Run tools in isolated, scalable environments
• Production Ready: Built-in error handling, retries, and monitoring
• Streamable HTTP: Uses FastMCP's streamable_http protocol for better performance
• Type Safety: Full type hints and validation

Quick Start

code architecture

aenv/ ├── pyproject.toml ├── src/ │ └── cli/ │ ├── init.py │ ├── cmds/ │ │ ├── init.py │ │ └── cli.py │ └── env/ │ ├── init.py │ └── another_file.py └── README.md

Installation

pip install aenvironment

Creating Tools

Create a Python file with your tools:

# my_tools.py
from aenv import register_tool

@register_tool
def get_weather(location: str, unit: str = "celsius") -> dict:
    """Get weather for a location."""
    return {"location": location, "temperature": 22.5, "unit": unit}

@register_tool
async def search_web(query: str, max_results: int = 10) -> list:
    """Search the web."""
    return [{"title": f"Result for {query}", "url": "https://example.com"}]

Local Development

Start an MCP server with your tools using FastMCP:

# Start MCP server on http://localhost:8081
python main.py ./my_tools.py

# Start with custom host and port
python main.py ./my_tools.py --host 0.0.0.0 --port 8080

# Start with tools from directory
python main.py ./tools/

# Start with custom server name
python main.py ./tools/ --name my-server

Using Environments

import asyncio
import Environment


async def main():
    # Create environment
    async with Environment("my-env") as environment:
        # List available tools
        tools = environment.list_tools()
        print("Available tools:", tools)

        # Call a tool
        result = await environment.call_tool(
            "my-env/get_weather",
            {"location": "Beijing", "unit": "celsius"}
        )
        print("Weather:", result.content)


asyncio.run(main())

Architecture

Core Concepts

• Tool: The smallest executable unit with input/output schemas
• Environment: Containerized runtime for tools with lifecycle management
• Registry: Global tool discovery and management

Directory Structure

aenv/
├── core/           # Core SDK components
├── server/         # FastMCP server implementation
├── client/         # API client for backend
└── examples/       # Usage examples

Tool Definition

Tools are defined using the @register_tool decorator:

from aenv import register_tool
from typing import Dict, Any

@register_tool
def my_tool(
    param1: str,
    param2: int = 42,
    param3: bool = False
) -> Dict[str, Any]:
    """
    Tool description for LLMs.

    Args:
        param1: Description of param1
        param2: Description of param2 (default: 42)
        param3: Description of param3

    Returns:
        Dictionary with results
    """
    return {"result": f"Processed {param1} with {param2}"}

Environment Management

Creating Environments

from aenv import Environment

# Create environment
env = Environment("my-env", scheduler_url="http://localhost:8080")
await env.initialize()

# Use as context manager
async with Environment("my-env") as env:
    tools = env.list_tools()
    result = await env.call_tool("tool_name", {"arg": "value"})

Configuration

Environment variables:

• AENV_SCHEDULER_URL: Default scheduler URL
• AENV_API_KEY: API key for authentication
• AENV_TIMEOUT: Default timeout in seconds

Error Handling

The SDK provides comprehensive error handling:

from env.core import (
    ToolError, ToolTimeoutError, EnvironmentError, NetworkError
)

try:
    result = await env.call_tool("tool_name", {"arg": "value"})
except ToolTimeoutError as e:
    print(f"Tool timed out: {e.timeout}s")
except ToolError as e:
    print(f"Tool failed: {e.message}")
except EnvironmentError as e:
    print(f"Environment error: {e.message}")

Testing

Run tests:

pip install -e ".[dev]"
pytest tests/

Examples

See the examples/ directory for complete examples:

• weather_tools.py: Weather-related tools
• search_tools.py: Web search and analysis tools

Development

Setup

git clone https://github.com/inclusionAI/AEnvironment
cd AEnvironment
pip install -e ".[dev]"

Running Tests

pytest tests/ -v

Code Quality

black aenv/
isort aenv/
mypy aenv/

License

MIT License - see LICENSE file for details.