AI Agent Hub
Back to plugins
dsh-eyes preview

dsh-eyes

Model Inference Updated 2026.08.25

Run the following command in DeepSeek Harness:

dsh plugin install Leeminjing/dsh-eyes

Paste the following prompt into your AI chat to install this plugin:

To install this plugin in DeepSeek Harness, run `dsh plugin install Leeminjing/dsh-eyes`, the source code is located at https://github.com/Leeminjing/dsh-eyes.

About this plugin

Pure-text models like DeepSeek often struggle with visual inputs, but the dsh-eyes plugin solves this by enabling text-only models to process images as smoothly as native multimodal ones. It bridges the gap between text and vision without manual intervention, creating a seamless workflow from "seeing" to "answering."

The core capability lies in its ease of use: users can simply paste an image via Ctrl+V or drag-and-drop attachments, which are then stored in the background. When the model needs to understand the image, it automatically invokes the view_image tool. Supporting any OpenAI-compatible vision endpoint and both Chat Completions and Responses APIs, it ensures high compatibility and silky-smooth interaction.

This plugin is ideal for developers and users looking to add vision capabilities to text-only models. Whether for OCR, UI analysis, or visual Q&A, dsh-eyes enhances efficiency and breaks the visual barrier of pure text models.

Screenshots

Use Cases

  • Image content recognition and OCR
  • UI layout analysis
  • Visual Q&A for text-only models

Best For

  • DeepSeek developers
  • AI workflow users needing vision assistance
  • DeepSeek Harness plugin users