ScriptsMar 29, 2026·1 min read

Smolagents — Lightweight AI Agent Framework

Hugging Face minimalist agent framework. Build AI agents in ~1000 lines of code with code-based actions, tool calling, and multi-step reasoning.

TO
TokRepo精选 · Community
Quick Use

Use it first, then decide how deep to go

This block should tell both the user and the agent what to copy, install, and apply first.

pip install smolagents

Intro

Smolagents is Hugging Face's lightweight agent framework — the core library is ~1000 lines of code. Agents write and execute Python code as their actions, making them more flexible and debuggable than JSON-based tool calling.

Best for: Quick agent prototyping, research, code-executing agents Works with: OpenAI, Anthropic, HuggingFace models, local models


Key Concept: Code Agents

Instead of outputting JSON tool calls, smolagents agents write Python code:

from smolagents import CodeAgent, tool, HfApiModel

@tool
def get_weather(city: str) -> str:
    """Get current weather for a city."""
    return f"Sunny, 22C in {city}"

agent = CodeAgent(tools=[get_weather], model=HfApiModel())
agent.run("What's the weather in Tokyo?")

The agent writes: result = get_weather("Tokyo") and executes it.


Why Code Agents?

  • More expressive — loops, conditionals, variables
  • Composable — chain tool calls naturally
  • Debuggable — read the generated code
  • Efficient — fewer LLM calls than ReAct

🙏

Source & Thanks

Created by Hugging Face. Licensed under Apache 2.0. huggingface/smolagents

Related Assets