Spaces:
Paused
Paused
|
Download README.md from tianyajum/python-agent: direct link, hf CLI and curl.
- Browser
- Download file 3.21 kB
-
https://huggingface.co/spaces/tianyajum/python-agent/resolve/main/README.md
- Command line
-
hf download hf://spaces/tianyajum/python-agent/README.md
-
curl -L -o README.md https://huggingface.co/spaces/tianyajum/python-agent/resolve/main/README.md
3.21 kB
A newer version of the Gradio SDK is available: 6.32.0
metadata
title: Python-Agent
emoji: 🤖
colorFrom: blue
colorTo: green
sdk: gradio
app_file: app.py
pinned: false
paused: true
Python Agent
A modular Python-based AI agent service built with FastAPI, integrating Nacos for service discovery and Model Context Protocol (MCP) for extensible tool usage. This agent uses LangGraph to orchestrate agentic workflows.
Features
- FastAPI Framework: High-performance, easy-to-learn web framework.
- Service Discovery: Seamless integration with Nacos for service registration and discovery.
- LangGraph: Stateful, multi-actor applications with cycles (agentic orchestration).
- Model Context Protocol (MCP):
- Stdio Client: Connects to local CLI tools (e.g., Brave Search via Node.js).
- SSE Client: Connects to remote services (e.g., Java backend) via Server-Sent Events.
Prerequisites
- Python 3.10+
- Node.js: Required for running the Brave Search MCP server (
npx). - Nacos Server 2.x: A running Nacos instance is required for the application to start.
Installation
Clone the repository:
git clone <repository_url> cd python-agentCreate and activate a virtual environment:
python -m venv .venv # Windows .venv\Scripts\activate # Linux/Mac source .venv/bin/activateInstall dependencies:
pip install -r requirements.txt
Configuration
The application is configured via environment variables. You can create a .env file in the root directory (see .env.example).
| Variable | Default | Description |
|---|---|---|
APP_ENV |
development |
Application environment (development or production). |
HOST |
0.0.0.0 |
Host to bind the server to. |
PORT |
8181 |
Port to listen on. |
NACOS_SERVER_ADDR |
`` | Address of the Nacos server. |
NACOS_NAMESPACE |
public |
Nacos namespace ID. |
NACOS_USERNAME |
`` | Nacos username. |
NACOS_PASSWORD |
`` | Nacos password. |
SERVICE_NAME |
python-agent |
Name of this service in Nacos. |
MCP_BRAVE_PATH |
(Auto-detected) | Optional override for npx path. |
NACOS_GATEWAY_SERVICE_NAME |
gateway |
Name of the Java/Gateway service to discover. |
Usage
Running the Application
Using python:
python main.py
Using uvicorn (with hot reload):
uvicorn main:app --reload --port 8181
Note: The application will fail to start if it cannot connect to the configured Nacos server.
API Endpoints
- Health Check:
GET /health - Chat:
POST /chat{ "message": "Search for the latest news on AI." }
Project Structure
The project follows a modular package structure:
python-agent/
├── app/
│ ├── core/ # Core configuration & infrastructure (Nacos, Config)
│ ├── services/ # External service clients (MCP)
│ ├── agent/ # LangGraph logic (State, Graph, Nodes)
│ └── setup.py # Startup setup scripts
├── main.py # Entry point
└── ...