Overview
dsh-mmroute
多模态路由 for DeepSeek Harness：纯文本模型的每一步请求里的每张图片（用户上传 / 工具返回 / MCP 渲染）都先经多模态理解模型转述为文字再作答，图片类报错自动转述重试，两类模型全程交叉协作。Multimodal router for DSH: every image in every step of a text-only model's stream is transcribed to text by a multimodal understander first; image-related failures auto-recover with a transcription retry.
README / EN
Package documentation
Registry summary
多模态路由 for DeepSeek Harness：纯文本模型的每一步请求里的每张图片（用户上传 / 工具返回 / MCP 渲染）都先经多模态理解模型转述为文字再作答，图片类报错自动转述重试，两类模型全程交叉协作。Multimodal router for DSH: every image in every step of a text-only model's stream is transcribed to text by a multimodal understander first; image-related failures auto-recover with a transcription retry.
dsh.pub verifies the pinned bundle contract, runtime facts, and distribution semantics. The complete README remains in the source repository.
Read the full README on GitHubLIMITATIONS
Known limitations
- 当前 DSH 附件系统 v1 仅支持**图片**（PNG / JPEG / WebP / GIF）；视频不在支持范围内。 - 理解模型的 token 消耗独立计费，不出现在主对话的用量统计中。 - 报错自愈依赖失败文案 / 错误码的图片特征启发式（`UNSUPPORTED_CONTENT` + image 字样，或 message 含 image / multimodal / vision / 视觉 / 图片）；无法识别的文案不会触发自愈，但显式「纯文本」标记仍会全程转述。