JoblessJoe/dsh-llm-ollama-native ↗★ 0
dsh-llm-ollama-native
DSH LLM adapter that speaks Ollama's native /api/chat instead of the OpenAI-compatible endpoint, so think:false/low/medium/high actually reaches the model. 适合需要将推理努力参数直接传递给Ollama模型的用户。
インストール
$
npx -p @deepseek-ai/dsh dsh plugin --profile web add github:JoblessJoe/dsh-llm-ollama-nativeドキュメント
README 全文を読む ↗Configuration reference
| Field | Required | Description |
|---|---|---|
provider | yes | Route name other dsh config refers to (e.g. agent-default-model.provider) |
baseURL | no | Ollama server address, default http://localhost:11434 |
displayName | no | Shown in model pickers, default Ollama (native) |
models | yes | Array of model entries |
models[].id | yes | Ollama model tag |
models[].contextWindow | no | Context window size, for display and request sizing |
models[].defaultReasoningEffort | no | off, low, medium, or high; used when a request specifies none, default low |