Movatterモバイル変換


[0]ホーム

URL:


Hugging Face's logoHugging Face

Hub documentation

Building with the SDK

Hub

Hugging Face's logo
Join the Hugging Face community

and get access to the augmented documentation experience

Collaborate on models, datasets and Spaces
Faster examples with accelerated inference
Switch between documentation themes

to get started

Building with the SDK

Build MCP-powered agents with the Hugging Face agentic SDKs. Thehuggingface_hub (Python) and@huggingface/tiny-agents (JavaScript) libraries provide everything you need to connect LLMs to MCP tools.

Installation

Python
JavaScript
pip install"huggingface_hub[mcp]"

Quick Start: Run an Agent

The fastest way to get started is with thetiny-agents CLI:

Python
JavaScript
tiny-agents run julien-c/flux-schnell-generator

This loads an agent from thetiny-agents collection, connects to its MCP servers, and starts an interactive chat.

Using the Agent Class

TheAgent class manages the chat loop and MCP tool execution. It usesInference Providers to run the LLM.

Python
JavaScript
from huggingface_hubimport Agentimport asyncioagent = Agent(    model="Qwen/Qwen2.5-72B-Instruct",    provider="novita",    servers=[        {"type":"sse","url":"https://evalstate-flux1-schnell.hf.space/gradio_api/mcp/sse"        }    ])asyncdefmain():asyncfor chunkin agent.run("Generate an image of a sunset"):ifhasattr(chunk,'choices'):            delta = chunk.choices[0].deltaif delta.content:print(delta.content, end="")asyncio.run(main())

See theAgent reference for all options.

Using MCPClient Directly

For more control, useMCPClient to manage MCP servers and tool calls directly.

Python
JavaScript
import asynciofrom huggingface_hubimport MCPClientasyncdefmain():asyncwith MCPClient(        model="Qwen/Qwen2.5-72B-Instruct",        provider="novita",    )as client:# Connect to an MCP serverawait client.add_mcp_server(type="sse",             url="https://evalstate-flux1-schnell.hf.space/gradio_api/mcp/sse"        )# Process a request with tools        messages = [{"role":"user","content":"Generate an image of a sunset"}]asyncfor chunkin client.process_single_turn_with_tools(messages):ifhasattr(chunk,'choices'):                delta = chunk.choices[0].deltaif delta.content:print(delta.content, end="")asyncio.run(main())

See theMCPClient reference for all options.

Share Your Agent

Contribute agents to thetiny-agents collection on the Hub. Include:

  • agent.json - Agent configuration (required)
  • PROMPT.md orAGENTS.md - System prompt (optional)
  • EXAMPLES.md - Sample prompts and use cases (optional)

Learn More

Update on GitHub


[8]ページ先頭

©2009-2026 Movatter.jp