chinese-llm-router
von xdd-xund
Coding Agents & IDEsKurzbeschreibung
Route your OpenClaw conversations to the best Chinese AI models — no config headaches, just pick and chat.
Beschreibung
Was dieser Skill macht
Routes OpenClaw conversations to 10 major Chinese LLMs through a single unified interface. Supports DeepSeek, Qwen, GLM, Kimi, Doubao, MiniMax, Step, Baichuan, Spark, and Hunyuan. Handles model switching, auto-fallback, cost tracking, and task-based recommendations.
Avoids maintaining separate API clients and configurations for each Chinese LLM provider by routing all requests through one interface with shared config and fallback logic.
Wann du ihn einsetzt
- Switching to DeepSeek R1 for a math-heavy problem
- Comparing Qwen and GLM outputs on the same coding task
- Summarizing a long Chinese document using Kimi's 128K context window
- Getting a model recommendation before starting a multi-step agent workflow
- Tracking token costs across multiple Chinese providers in one session
Beispiel-Workflow
User asks: I need to debug a complex Python script. Which model should I use and can you switch to it?
- 1
Identifies the task as a coding request
- 2
Consults the model selection guide and recommends GLM-5 or Kimi K2.5 for coding
- 3
Switches the active model to GLM-5
- 4
Confirms the switch and passes the Python script to GLM-5 for debugging
GLM-5 activated; debugging analysis returned for the Python script
Voraussetzungen
- DeepSeek API key (platform.deepseek.com)
- Qwen API key via DashScope (dashscope.console.aliyun.com)
- GLM API key via Zhipu AI (open.bigmodel.cn)
- Kimi API key via Moonshot AI (platform.moonshot.cn)
- Doubao API key via VolcEngine (console.volcengine.com/ark)
- MiniMax API key (platform.minimaxi.com)
- StepFun API key (platform.stepfun.com)
- Baichuan API key (platform.baichuan-ai.com)
- Spark API key via iFlytek (console.xfyun.cn)
- Hunyuan API key via Tencent Cloud (cloud.tencent.com/product/hunyuan)
Installation
clawhub install xdd-xund/chinese-llm-routerFühre den CLI-Befehl aus oder füge den Prompt in Claude Code, Cursor, Codex oder deinen Coding-Agenten ein.