llm-adaptive
Adaptive model routing for DeepSeek Harness: per-request complexity classification with automatic provider routing.
AI Analysis
核心用途是根据用户输入的复杂度（低/中/高/关键）自动分流并路由至最合适的模型。适合需要平衡响应速度、模型能力与 API 成本的用户。
Install
$
npx -p @deepseek-ai/dsh dsh plugin --profile web add github:dylan121322/llm-adaptiveREADME
Read the full README ↗Usage
- Open
/modeland chooseadaptive（自动路由）. - Every subsequent LLM request is classified (low/medium/high/critical) and routed to the first available provider of that level's chain.
- Decisions are logged with
level=… cause=… chain=…to~/.dsh/hooks/plugin.log.
The explicit level models (low, medium, high, critical) are also listed in the picker for direct selection.