Bazley13/dsh-multi-model-orchestrator ↗★ 1
dsh-multi-model-orchestrator
A DeepSeek Harness (dsh) plugin: orchestrates complex tasks across multiple models — a 'main brain' decomposes the task and dispatches subtasks to GLM / Kimi / Qwen etc. sub-agents by each model's strengths, with per-model token usage.
AI 分析
核心用途是多模型协同任务编排，将复杂任务拆解并分发给最擅长的模型。适合需要混合使用多种国产/外置大模型进行复杂工作流处理的开发者和高级用户。
インストール
$
npx -p @deepseek-ai/dsh dsh plugin --profile web add github:Bazley13/dsh-multi-model-orchestratorドキュメント
README 全文を読む ↗Configuration reference
Everything is configured in $DSH_HOME/settings.yaml:
| Section | Purpose |
|---|---|
llm-pi-ai.providers | OpenAI-compatible third-party model routes (any vendor). |
subagent-model-selection | Allow-list of {provider, model} the sub-agent tool may dispatch to. |
multi-model-orchestrator.modelNotes | Per-route strengths/weaknesses that guide assignment. |
To add your own vendor, extend llm-pi-ai.providers with { api, baseURL, apiKeyEnv, models } (any OpenAI-compatible api: openai-completions gateway works), then add matching entries to subagent-model-selection.allowedModels and multi-model-orchestrator.modelNotes.