一个简单的智能体应用,支持调用tools和MCP的能力,更多功能可以参考代码自行实现
Agent框架:LangGraph
安装依赖
pip install uv uv sync
配置.env文件中的环境变量,运行run.py启动服务,访问http://127.0.0.1:8080页面即可开始聊天
make
make linux-amd64
docker save think-probe:1.0.0 | gzip > think-probe-1.0.0.tar.gz
docker load < think-probe-1.0.0.tar.gzdocker run --name think-probe -e LLM_API_PATH=http://192.168.31.26:1234/v1 -p 18080:8080 -d think-probe:1.0.0
创建docker-compose.yml文件,修改相关环境变量:
services:
think-probe:
container_name: think-probe
image: think-probe:1.0.0
ports:
- "18080:8080"
environment:
- "OPENAI_API_KEY=sk-xxx"
- "LLM_API_PATH=https://openrouter.ai/api/v1"
- "LLM_API_KEY=sk-xxx"
- "LLM_MODEL_NAME=qwen/qwen-2.5-72b-instruct"
networks:
- net-think-probe
networks:
net-think-probe:
driver: bridge
保存后执行
docker-compose up -d