A utility for reading configurations from YAML files to run Langchain workflows.
- 🛠️ YAML-based configuration management
- ⛓️ Langchain integration
- 🌐 OpenAI model support
- 📦 Pydantic data validation
pip install langchain-roleplay- 创建配置文件
test.role.ymal：
# 示例配置
name: 呆猫
prompt:
- role: system
content: |
你是呆猫，尊称用户为老大。
你需要适时且克制地在输出的语句中插入‘喵’。
尽量输出口语化的语句。
- role: placeholder
content: '{history}'
memory:
type: base
llm: deepseek
config:
model: deepseek-chat
max_tokens: 8192
temperature: 0.65
as_tool:
name: 询问呆猫
parameters:
- name: input
type: str
description: 输入
description: 让呆猫回答你的问题
tools:
- get_weather- 创建llm配置文件
test.llm.yaml
name: deepseek
base_url: 'https://api.deepseek.com'
api_key: sk-xxx # 替换为你的API Key- 创建工具配置文件
test.tool.yaml：
name: 'get_weather'
parameters:
- name: 'location'
type: 'str'
description: 'The location to get weather'
description: 'Get weather of an location, the user shoud supply a location first'
script: './py/test.py' # 自行实现- 在Python中使用：
from roleplay.core import load_llm, load_tool
from roleplay.memory import load_role
load_llm('test.llm.yaml')
load_tool('test.tool.yaml')
role = load_role('test.role.yaml')
result = role.run(input='你好')
print(result)- Python 3.8+
- langchain >= 0.3.17
- langgraph >= 0.2.70
- pydantic >= 2.10.6
- PyYAML >= 6.0.2
- langchain-openai >= 0.3.4
git clone https://github.com/maye76/roleplay.git
cd roleplay
pip install -e .欢迎提交Issue和PR！
MIT © 2025 maye76