Settings

本例展示全局设置、实例覆盖与 debug 开关。

源码

from agently import Agently

## Settings
# Global Model Settings
Agently.set_settings(
    "OpenAICompatible",
    {
        "base_url": "http://127.0.0.1:11434/v1",
        "model": "qwen2.5:7b",
    },
)

# Create LLM Request Agent Instance
agent = Agently.create_agent()

# Keys' values in agent instance settings will cover global settings
# but not keys' values that not mention will inherit from global settings
agent.set_settings(
    "OpenAICompatible",
    {
        "model": "qwen3:latest",
    },
)

## Debug Toggle
# Set to False by default, turn on debug mode will display model request processing streaming logs in console.
agent.set_settings("debug", True)

agent_model_requester_settings = agent.settings.get("plugins.ModelRequester.OpenAICompatible", {})
assert isinstance(agent_model_requester_settings, dict)
print(agent_model_requester_settings.get("base_url"))  # "http://127.0.0.1:11434/v1"
print(agent_model_requester_settings.get("model"))  # qwen3:latest

## Note:
# Default Global Settings: agently/_default_settings.yaml
# Default Plugin Settings can be defined in attribution "DEFAULT_SETTINGS" in Plugin Class
# Core Plugins with Settings:
# Model Requester: agently/builtins/plugins/ModelRequester/OpenAICompatible.py

讲解

• 全局设置提供默认值。
• 实例设置覆盖指定字段。
• debug 会联动多项运行时日志。

注释解读

• Idea 表示案例思路
• Flow 表示执行编排路径
• Expect 表示预期输出或行为

你学会了什么

• 理解层级设置的优先级与覆盖规则
• 掌握 debug 与运行时日志的联动
• 知道如何按需修改模型配置

练习任务

• 把 model 从 qwen2.5 改成你本地的另一个模型
• 切换 debug 并观察日志变化