澄潭
澄潭
## Why do you need it? When using the streaming interface of deepseek-r1 from dashscope, reasoningContent also needs to be output in a streaming manner; otherwise, the API will wait...
基于 LLM Mock Server 增强 ai-proxy 插件的 e2e 测试进度如下 ### 目前已支持 - baidu - doubao - minimax ### 待支持 #### 因响应格式与OpenAI不同,需在 Mock Server 中实现对应响应 - [ ] #1717 - [...
## Why do you need it? Because running model services in an e2e environment or accessing the LLM provider's API is not very feasible. The e2e testing of plugins for...
代理字节扣子相关api,实现对应llm mock server逻辑 https://www.coze.cn/open/docs/developer_guides/chat_v3
## Why do you need it? 之前的 cors 插件实现不够严谨,具体如下: 1. 在判断是否要触发对预检请求直接生成响应时,没有检查 Access-Control-Request-* 相关头; 2. 对于携带 Origin 请求头,但是 method/headers 等和配置不同的请求,直接返回了 403, 这个没有必要,因为浏览器会做相应的判断: https://github.com/alibaba/higress/blob/b997e6fd265aaaf5b9de6a8d44acb2a3079fa05e/plugins/wasm-go/extensions/cors/config/cors_config.go#L281-L310 3. 应该增加 Vary: Origin 响应头,来避免浏览器缓存导致跨域失效