smolagents｜用代码执行动作的轻量 Python Agent 框架

access：来源确认 · pip install "smolagents[toolkit]"

project_type：来源确认 · `smolagents` is a library that enables you to run powerful agents in a few lines of code.

<!---
Copyright 2024 The HuggingFace Team. All rights reserved.

Licensed under the Apache License, Version 2.0 (the "License");
you may not use this file except in compliance with the License.
You may obtain a copy of the License at

    http://www.apache.org/licenses/LICENSE-2.0

Unless required by applicable law or agreed to in writing, software
distributed under the License is distributed on an "AS IS" BASIS,
WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
See the License for the specific language governing permissions and
limitations under the License.
-->
<p align="center">
    <!-- Uncomment when CircleCI is set up
    <a href="https://circleci.com/gh/huggingface/accelerate"><img alt="Build" src="https://img.shields.io/circleci/build/github/huggingface/transformers/master"></a>
    -->
    <a href="https://github.com/huggingface/smolagents/blob/main/LICENSE"><img alt="License" src="https://img.shields.io/github/license/huggingface/smolagents.svg?color=blue"></a>
    <a href="https://huggingface.co/docs/smolagents"><img alt="Documentation" src="https://img.shields.io/website/http/huggingface.co/docs/smolagents/index.html.svg?down_color=red&down_message=offline&up_message=online"></a>
    <a href="https://github.com/huggingface/smolagents/releases"><img alt="GitHub release" src="https://img.shields.io/github/release/huggingface/smolagents.svg"></a>
    <a href="https://github.com/huggingface/smolagents/blob/main/CODE_OF_CONDUCT.md"><img alt="Contributor Covenant" src="https://img.shields.io/badge/Contributor%20Covenant-v2.0%20adopted-ff69b4.svg"></a>
    <a href="https://deepwiki.com/huggingface/smolagents"><img src="https://deepwiki.com/badge.svg" alt="Ask DeepWiki"></a>
</p>

<h3 align="center">
  <div style="display:flex;flex-direction:row;">
    <img src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/smolagents/smolagents.png" alt="Hugging Face mascot as James Bond" width=400px>
    <p>Agents that think in code!</p>
  </div>
</h3>

`smolagents` is a library that enables you to run powerful agents in a few lines of code. It offers:

✨ **Simplicity**: the logic for agents fits in ~1,000 lines of code (see [agents.py](https://github.com/huggingface/smolagents/blob/main/src/smolagents/agents.py)). We kept abstractions to their minimal shape above raw code!

🧑‍💻 **First-class support for Code Agents**. Our [`CodeAgent`](https://huggingface.co/docs/smolagents/reference/agents#smolagents.CodeAgent) writes its actions in code (as opposed to "agents being used to write code"). To make it secure, we support executing in sandboxed environments via [Blaxel](https://blaxel.ai), [E2B](https://e2b.dev/), [Modal](https://modal.com/), or Docker.

🤗 **Hub integrations**: you can [share/pull tools or agents to/from the Hub](https://huggingface.co/docs/smolagents/reference/tools#smolagents.Tool.from_hub) for instant sharing of the most efficient agents!

🌐 **Model-agnostic**: smolagents supports any LLM. It can be a local `transformers` or `ollama` model, one of [many providers on the Hub](https://huggingface.co/blog/inference-providers), or any model from OpenAI, Anthropic and many others via our [LiteLLM](https://www.litellm.ai/) integration.

👁️ **Modality-agnostic**: Agents support text, vision, video, even audio inputs! Cf [this tutorial](https://huggingface.co/docs/smolagents/examples/web_browser) for vision.

🛠️ **Tool-agnostic**: you can use tools from any [MCP server](https://huggingface.co/docs/smolagents/reference/tools#smolagents.ToolCollection.from_mcp), from [LangChain](https://huggingface.co/docs/smolagents/reference/tools#smolagents.Tool.from_langchain), you can even use a [Hub Space](https://huggingface.co/docs/smolagents/reference/tools#smolagents.Tool.from_space) as a tool.

Full documentation can be found [here](https://huggingface.co/docs/smolagents/index).

> [!NOTE]
> Check out our [launch blog post](https://huggingface.co/blog/smolagents) to learn more about `smolagents`!

## Quick demo

First install the package with a default set of tools:
```bash
pip install "smolagents[toolkit]"
```
Then define your agent, give it the tools it needs and run it!
```py
from smolagents import CodeAgent, WebSearchTool, InferenceClientModel

model = InferenceClientModel()
agent = CodeAgent(tools=[WebSearchTool()], model=model, stream_outputs=True)

agent.run("How many seconds would it take for a leopard at full speed to run through Pont des Arts?")
```

https://github.com/user-attachments/assets/84b149b4-246c-40c9-a48d-ba013b08e600

You can even share your agent to the Hub, as a Space repository:
```py
agent.push_to_hub("m-ric/my_agent")

# agent.from_hub("m-ric/my_agent") to load an agent from Hub
```

Our library is LLM-agnostic: you could switch the example above to any inference provider.

<details>
<summary> <b>InferenceClientModel, gateway for all <a href="https://huggingface.co/docs/inference-providers/index">inference providers</a> supported on HF</b></summary>

```py
from smolagents import InferenceClientModel

model = InferenceClientModel(
    model_id="deepseek-ai/DeepSeek-R1",
    provider="together",
)
```
</details>
<details>
<summary> <b>LiteLLM to access 100+ LLMs</b></summary>

```py
from smolagents import LiteLLMModel

model = LiteLLMModel(
    model_id="anthropic/claude-4-sonnet-latest",
    temperature=0.2,
    api_key=os.environ["ANTHROPIC_API_KEY"]
)
```
</details>
<details>
<summary> <b>OpenAI-compatible servers: Together AI</b></summary>

```py
import os
from smolagents import OpenAIModel

model = OpenAIModel(
    model_id="deepseek-ai/DeepSeek-R1",
    api_base="https://api.together.xyz/v1/", # Leave this blank to query OpenAI servers.
    api_key=os.environ["TOGETHER_API_KEY"], # Switch to the API key for the server you're targeting.
)
```
</details>
<details>
<summary> <b>OpenAI-compatible servers: OpenRouter</b></summary>

```py
import os
from smolagents import OpenAIModel

model = OpenAIModel(
    model_id="openai/gpt-4o",
    api_base="https://openrouter.ai/api/v1", # Leave this blank to query OpenAI servers.
    api_key=os.environ["OPENROUTER_API_KEY"], # Switch to the API key for the server you're targeting.
)
```

</details>
<details>
<summary> <b>Local `transformers` model</b></summary>

```py
from smolagents import TransformersModel

model = TransformersModel(
    model_id="Qwen/Qwen3-Next-80B-A3B-Thinking",
    max_new_tokens=4096,
    device_map="auto"
)
```
</details>
<details>
<summary> <b>Azure models</b></summary>

```py
import os
from smolagents import AzureOpenAIModel

model = AzureOpenAIModel(
    model_id = os.environ.get("AZURE_OPENAI_MODEL"),
    azure_endpoint=os.environ.get("AZURE_OPENAI_ENDPOINT"),
    api_key=os.environ.get("AZURE_OPENAI_API_KEY"),
    api_version=os.environ.get("OPENAI_API_VERSION")    
)
```
</details>
<details>
<summary> <b>Amazon Bedrock models</b></summary>

```py
import os
from smolagents import AmazonBedrockModel

model = AmazonBedrockModel(
    model_id = os.environ.get("AMAZON_BEDROCK_MODEL_ID") 
)
```
</details>

## CLI

You can run agents from CLI using two commands: `smolagent` and `webagent`.

`smolagent` is a generalist command to run a multi-step `CodeAgent` that can be equipped with various tools.

```bash
# Run with direct prompt and options
smolagent "Plan a trip to Tokyo, Kyoto and Osaka between Mar 28 and Apr 7."  --model-type "InferenceClientModel" --model-id "Qwen/Qwen3-Next-80B-A3B-Thinking" --imports pandas numpy --tools web_search

# Run in interactive mode (launches setup wizard when no prompt provided)
smolagent
```

Interactive mode guides you through:
- Agent type selection (CodeAgent vs ToolCallingAgent)  
- Tool selection from available toolbox
- Model configuration (type, ID, API settings)
- Advanced options like additional imports
- Task prompt input

Meanwhile `webagent` is a specific web-browsing agent using [helium](https://github.com/mherrmann/helium) (read more [here](https://github.com/huggingface/smolagents/blob/main/src/smolagents/vision_web_browser.py)).

For instance:
```bash
webagent "go to xyz.com/men, get to sale section, click the first clothing item you see. Get the product details, and the price, return them. note that I'm shopping from France" --model-type "LiteLLMModel" --model-id "gpt-5"
```

## How do Code agents work?

Our [`CodeAgent`](https://huggingface.co/docs/smolagents/reference/agents#smolagents.CodeAgent) works mostly like classical ReAct agents - the exception being that the LLM engine writes its actions as Python code snippets.

```mermaid
flowchart TB
    Task[User Task]
    Memory[agent.memory]
    Generate[Generate from agent.model]
    Execute[Execute Code action - Tool calls are written as functions]
    Answer[Return the argument given to 'final_answer']

    Task -->|Add task to agent.memory| Memory

    subgraph ReAct[ReAct loop]
        Memory -->|Memory as chat messages| Generate
        Generate -->|Parse output to extract code action| Execute
        Execute -->|No call to 'final_answer' tool => Store execution logs in memory and keep running| Memory
    end
    
    Execute -->|Call to 'final_answer' tool| Answer

    %% Styling
    classDef default fill:#d4b702,stroke:#8b7701,color:#ffffff
    classDef io fill:#4a5568,stroke:#2d3748,color:#ffffff
    
    class Task,Answer io
```

Actions are now Python code snippets. Hence, tool calls will be performed as Python function calls. For instance, here is how the agent can perform web search over several websites in one single action:
```py
requests_to_search = ["gulf of mexico america", "greenland denmark", "tariffs"]
for request in requests_to_search:
    print(f"Here are the search results for {request}:", web_search(request))
```

Writing actions as code snippets is demonstrated to work better than the current industry practice of letting the LLM output a dictionary of the tools it wants to call: [uses 30% fewer steps](https://huggingface.co/papers/2402.01030) (thus 30% fewer LLM calls) and [reaches higher performance on difficult benchmarks](https://huggingface.co/papers/2411.01747). Head to [our high-level intro to agents](https://huggingface.co/docs/smolagents/conceptual_guides/intro_agents) to learn more on that.

Since code execution can be a serious security concern (arbitrary code execution!), **you should run agent code in a sandbox**. We support several options:
  - [E2B](https://e2b.dev/), [Blaxel](https://blaxel.ai), [Modal](https://modal.com/) — managed cloud sandboxes, simplest to set up
  - [Docker](https://www.docker.com/) — self-hosted container isolation

The built-in `LocalPythonExecutor` is **not a security sandbox**. It applies some restrictions but can be bypassed and must not be used as a security boundary.

Alongside [`CodeAgent`](https://huggingface.co/docs/smolagents/reference/agents#smolagents.CodeAgent), we also provide the standard [`ToolCallingAgent`](https://huggingface.co/docs/smolagents/reference/agents#smolagents.ToolCallingAgent) which writes actions as JSON/text blobs. You can pick whichever style best suits your use case.

## How smol is this library?

We strived to keep abstractions to a strict minimum: the main code in `agents.py` has <1,000 lines of code.
Still, we implement several types of agents: `CodeAgent` writes its actions as Python code snippets, and the more classic `ToolCallingAgent` leverages built-in tool calling methods. We also have multi-agent hierarchies, import from tool collections, remote code execution, vision models...

By the way, why use a framework at all? Well, because a big part of this stuff is non-trivial. For instance, the code agent has to keep a consistent format for code throughout its system prompt, its parser, the execution. So our framework handles this complexity for you. But of course we still encourage you to hack into the source code and use only the bits that you need, to the exclusion of everything else!

## How strong are open models for agentic workflows?

We've created [`CodeAgent`](https://huggingface.co/docs/smolagents/reference/agents#smolagents.CodeAgent) instances with some leading models, and compared them on [this benchmark](https://huggingface.co/datasets/m-ric/agents_medium_benchmark_2) that gathers questions from a few different benchmarks to propose a varied blend of challenges.

[Find the benchmarking code here](https://github.com/huggingface/smolagents/blob/main/examples/smolagents_benchmark/run.py) for more detail on the agentic setup used, and see a comparison of using LLMs code agents compared to vanilla (spoilers: code agents works better).

<p align="center">
    <img src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/smolagents/benchmark_code_agents.jpeg" alt="benchmark of different models on agentic workflows. Open model DeepSeek-R1 beats closed-source models." width=60% max-width=500px>
</p>

This comparison shows that open-source models can now take on the best closed models!

## Security

Security is a critical consideration when working with code-executing agents. Ensure you are using one of the sandboxed execution options that provide isolation from untrusted code.

**Warning:** `LocalPythonExecutor` provides best-effort mitigations only and is **not a security boundary**. Do not use it to run untrusted code.

For security policies, vulnerability reporting, and more information on secure agent execution, please see our [Security Policy](SECURITY.md).

## Contribute

Everyone is welcome to contribute, get started with our [contribution guide](https://github.com/huggingface/smolagents/blob/main/CONTRIBUTING.md).

## Cite smolagents

If you use `smolagents` in your publication, please cite it by using the following BibTeX entry.

```bibtex
@Misc{smolagents,
  title =        {`smolagents`: a smol library to build great agentic systems.},
  author =       {Aymeric Roucher and Albert Villanova del Moral and Thomas Wolf and Leandro von Werra and Erik Kaunismäki},
  howpublished = {\url{https://github.com/huggingface/smolagents}},
  year =         {2025}
}
```

需要 Python 环境与 pip；按 README 使用 pip install "smolagents[toolkit]" 安装。使用云端模型需相应服务商 API Key（如 ANTHROPIC_API_KEY、TOGETHER_API_KEY、OPENROUTER_API_KEY、AZURE_OPENAI_* 等）；本地 transformers 模型需足够算力。云模型、E2B/Blaxel/Modal 沙箱及 Hub 服务的费用待向各官方核验。支持的操作系统与国内网络可用性待核验。