Add TypeScript clients for GLM4.7 and Qwen3 models

[Copilot]

Ports GLM4.7 and Qwen3 client implementations from Python to TypeScript.

Implementation

• New clients (src_ts/src/glm4_7/, src_ts/src/qwen3/)

  • Both use OpenAI-compatible API via OpenAI SDK
  • GLM4.7: Supports thinking configuration, tool streaming via extra_body
  • Qwen3: Handles two tool call formats (streamed JSON and delta-based)

• AutoClient routing (src_ts/src/autoClient.ts)

  • Added client detection for glm-4.7 and qwen3 model identifiers
  • Removed placeholder error messages

• Test integration (src_ts/tests/client.test.ts)

  • Models auto-included when GLM_API_KEY environment variable is set for direct GLM4.7 testing
  • Added OpenRouter support: z-ai/glm-4.7 and qwen/qwen3-30b-a3b-thinking-2507 when OPENROUTER_API_KEY is set
  • Added SiliconFlow support: Pro/zai-org/GLM-4.7 and Qwen/Qwen3-8B when SILICONFLOW_API_KEY is set
  • Test structure matches Python implementation with automatic API key and base URL selection
  • Separated text-only models from vision models

Example Usage

import { AutoLLMClient } from "agenthub-typescript";

// Direct usage with native API
const glmClient = new AutoLLMClient("glm-4.7");

// OpenRouter usage
const openrouterClient = new AutoLLMClient(
  "z-ai/glm-4.7",
  process.env.OPENROUTER_API_KEY,
  "https://openrouter.ai/api/v1"
);

// SiliconFlow usage
const siliconflowClient = new AutoLLMClient(
  "Qwen/Qwen3-8B",
  process.env.SILICONFLOW_API_KEY,
  "https://api.siliconflow.cn/v1"
);

for await (const event of glmClient.streamingResponse(messages, config)) {
  // Handle streaming events
}

Both implementations match their Python counterparts in src_py/agenthub/glm4_7/ and src_py/agenthub/qwen3/.

Original prompt

基于src_py里面python的glm4.7和Qwen3的实现，在src_ts里面添加typescript版本的client，记得更新ts的test

---

✨ Let Copilot coding agent set things up for you — coding agent works faster and does higher quality work when set up for your repo.