DSH Vision Recognizer
kaixinbaba/dsh-vision-recognizer · model-adapters
The wrapper always admits image attachments, then resolves the exact selected model: models declaring native image input receive the original image blocks directly
★ 2GitHub stars
0Forks
2026-08-25Last updated
JavaScriptLanguage
MITLicense
Key features
- Adaptive routingnative multimodal models receive images unchanged
- Keep DeepSeek as the conversation brain, attach images anyway, and switch the image-recognition provider any time from Settings → Plugins.
- ctx.llm.listModels(innerProvider) and resolveModelInfo(innerProvider, model) inspect the exact target model and rebind its metadata to the wrapper route
- only text-only or unknown-capability models invoke the configured transcription fallback.
- One-click installdsh plugin --profile web add dsh-vision-recognizer — no build scripts, no sharp approval (no native dependencies at all).
Requirements
- Capability lookup and prepared target dispatch are separate public operations in rc8.
- A target adapter replaced by HMR in that tiny interval can race the routing decision.
Install command
dsh plugin --profile web add github:kaixinbaba/dsh-vision-recognizer