DeepSeek HarnessHub
提交插件

bamboostrip/dsh-llm-capabilities ↗★ 0

dsh-llm-capabilities

DSH 模型能力自动检测与配置插件，支持检测推理努力程度（thinking levels）及输入模态并应用至设置。

AI 分析

核心用途是精细化模型参数配置。适合需要针对不同模型微调思考深度、视觉模态等高级特性的专业用户。

視覺
套件
dsh-llm-capabilities
版本
0.1.0
授權
MIT
最近更新
2026年8月26日

安裝

$npx -p @deepseek-ai/dsh dsh plugin --profile web add github:bamboostrip/dsh-llm-capabilities

說明文件

閱讀完整 README ↗

Usage

  1. Add your provider in Settings → Models as usual (route, baseURL, apiKeyEnv).
  2. Open Settings → Model Capabilities, pick the provider.
  3. Click Detect from endpoint (or Pre-fill from catalog for catalog models).
  4. Toggle per model:
    • Offer thinking levels → pick off/low/medium/high/xhigh/max, edit wire spellings inline
    • Vision: inherit / image ✓ / text-only
  5. Apply to settings → writes to llm-pi-ai (providers..models). Intensity slider and image admission update immediately.

Configuration shape written

llm-pi-ai:
  providers:
    my-gateway:
      baseURL: https://gateway.example/v1
      api: openai-completions
      models:
        - id: gpt-4o
          input: [text, image]          # vision
          reasoningEfforts:              # thinking
            off: null
            low: low
            medium: medium
            high: high
        - id: text-only-model
          input: [text]
          reasoningEfforts: false        # explicitly no reasoning