Buckets:

download
raw
2.64 kB
"""Plugin example — create a custom tool plugin and use it in an agent.
Demonstrates how to:
1. Define a custom tool by subclassing BaseTool
2. Package it as a ToolPlugin
3. Register it with PluginManager
4. Use it in an agent
Usage:
CUDA_VISIBLE_DEVICES=0 python examples/plugin_example.py
"""
import logging
import os
os.environ.setdefault("CUDA_VISIBLE_DEVICES", "0")
logging.basicConfig(level=logging.WARNING, format="%(levelname)s: %(message)s")
from effgen import Agent, AgentConfig, load_model
from effgen.tools.base_tool import (
BaseTool,
ParameterSpec,
ParameterType,
ToolCategory,
ToolMetadata,
ToolResult,
)
from effgen.tools.plugin import PluginManager, ToolPlugin
# Step 1: Define a custom tool
class GreetingTool(BaseTool):
"""A simple tool that generates greetings."""
def __init__(self):
super().__init__(
metadata=ToolMetadata(
name="greeting",
description="Generate a friendly greeting for a given name.",
category=ToolCategory.COMMUNICATION,
parameters=[
ParameterSpec(
name="name",
type=ParameterType.STRING,
description="The name to greet",
required=True,
),
],
)
)
async def _execute(self, name: str, **kwargs) -> ToolResult:
greeting = f"Hello, {name}! Welcome to effGen!"
return ToolResult(success=True, output=greeting)
# Step 2: Create a plugin containing the tool
class GreetingPlugin(ToolPlugin):
name = "greeting_plugin"
version = "1.0.0"
description = "A demo plugin that adds a greeting tool."
tools = [GreetingTool]
# Step 3: Register the plugin
manager = PluginManager()
plugin = GreetingPlugin()
registered = plugin.register(manager.registry)
print(f"Registered tools: {registered}")
# Step 4: Use the custom tool in an agent
MODEL_ID = "Qwen/Qwen2.5-3B-Instruct"
print(f"\nLoading model: {MODEL_ID}")
model = load_model(MODEL_ID, quantization="4bit")
config = AgentConfig(
name="plugin_agent",
model=model,
tools=[GreetingTool()],
system_prompt="You are a friendly assistant. Use the greeting tool to greet people.",
max_iterations=5,
)
agent = Agent(config=config)
print("\n" + "=" * 60)
print("Query: 'Please greet Alice'")
print("=" * 60)
result = agent.run("Please greet Alice.")
print(f"\nAnswer: {result.output}")
print(f"Success: {result.success}")
print("\n" + "=" * 60)
print("Plugin example complete!")
print("=" * 60)

Xet Storage Details

Size:
2.64 kB
·
Xet hash:
08a63e8b6b4c2de284bee938a8cb2d552adc2b468ac1d32f709d001c8c7445af

Xet efficiently stores files, intelligently splitting them into unique chunks and accelerating uploads and downloads. More info.