JuneLearn/dsh-reasoning-settings ↗★ 5
dsh-reasoning-settings
适用于 DeepSeek Harness 自定义服务商的推理强度设置与单次调用子智能体模型路由插件。
AI 分析
核心用途是让用户能为自定义模型配置和选择不同的推理强度（Reasoning effort）。适合使用支持推理强度调节的自定义 API 服务商，并希望在 DSH 模型选择器中自由切换推理等级的用户。
安裝
$
npx -p @deepseek-ai/dsh dsh plugin --profile web add github:JuneLearn/dsh-reasoning-settings說明文件
閱讀完整 README ↗Usage
- Add your custom provider and models on the official Models page first.
- Open Settings > Reasoning effort.
- Pick the supported levels for each model and set the provider default effort.
- Click Save under that provider.
- Start a new session and choose the model and reasoning effort in the model picker.
The plugin only declares which effort levels Harness may select and send. Whether the relay API actually honors a value is up to the relay: if a request comes back with HTTP 400, uncheck the levels that model does not support or adjust their wire values.