Skillsdsh-plugin

dsh-eyes

Give text-only DeepSeek models on-demand vision: upload images, DeepSeek answers by calling a view_image tool backed by any OpenAI-compatible vision endpoint (Qwen/DashScope by default).

Stars
1
Forks
0
Open issues
0
Last push
Aug 16, 2026
Latest release
v0.1.0
24h Growth
+0
7d Growth
+0

Installation

dsh plugin --profile web add github:Leeminjing/dsh-eyes

Overview

Give text-only DeepSeek models on-demand vision: upload images, DeepSeek answers by calling a view_image tool backed by any OpenAI-compatible vision endpoint (Qwen/DashScope by default).

  • Bypass image send restriction: text-only DeepSeek models accept and retain image attachments in chat.
  • Images are replaced with reference notes (【图片N attachment_id=…】) before dispatch, so the main model gets text.
  • Registers a view_image tool the main model can call on demand to inspect a stored image via a vision API.
  • Works with any OpenAI-compatible vision endpoint; defaults to Alibaba Cloud Model Studio (Qwen).
  • Supports Chat Completions and Responses API styles with automatic detection or manual override.
  • Automatically adapts to any text-only main model; native multimodal models remain unaffected.
  • Attachment index is session-scoped and persisted to .dsh/attachments, surviving context compression or restarts.
  • Can view multiple images at once (attachment_ids) or from a local image_path (up to 15 MB).

Related plugins