dsh-pro-vision-std
Run the following command in DeepSeek Harness:
dsh plugin install lasdrder0705/dsh-pro-vision-std
Paste the following prompt into your AI chat to install this plugin:
In DeepSeek Harness, visit the source at https://github.com/lasdrder0705/dsh-pro-vision-std and run dsh plugin install lasdrder0705/dsh-pro-vision-std to install this plugin.
About this plugin
DeepSeek V4-Pro is a text-only inference model, and the official route does not accept image attachments. Yet in everyday work, screenshots, charts, and UI mockups keep showing up in conversations. dsh-pro-vision-std fills that gap: it upgrades the V4-Pro route to handle images without changing your existing workflow.
The mechanism is straightforward. Each image is first captioned by deepseek-v4-flash-vision-exp into a precise text description, which is embedded in the message as an [Image N] note. V4-Pro then streams its final reply from the rewritten, text-only request. Identical image bytes are cached in-process, so pasting the same image twice will not trigger a second vision call. The plugin also exposes the dsh-std describe_image tool for explicitly describing a single workspace or recently attached image. API keys are resolved automatically through the DSH credential service, so there is no need to re-enter them. The whole component is built on the dsh-std Community v0.15 contract, has zero dependency on @deepseek-ai packages, and coexists as an isolated route alongside the official text-only route.
If you already use V4-Pro in DSH for coding or analysis and keep needing it to take a look at an image, this is the lightest bridge you can install. It is also a good fit for dsh-std ecosystem developers who want a portable, standards-based ModelProvider that behaves the same across hosts and deployment targets.
Use Cases
- Paste a screenshot or chart into a V4-Pro conversation to get visual-aware analysis
- Have V4-Pro read a UI mockup or error screenshot and return targeted suggestions
- Bridge vision capability through standard dsh-std contracts across hosts
Best For
- Developers already using V4-Pro for code analysis or writing
- Engineers who need image understanding without switching models
- Ecosystem developers seeking portable, standards-based inference plugins
Related Plugins
ModLens is a vision plugin for DeepSeek Harness that gives text-only models sight by reading images pasted directly into chat, with zero-config setup and multiple vision engines.
On-demand vision for text-only DeepSeek Harness agents: built-in free keyless vision chain and 14 vision tools, routing image turns as tool calls to vision models with pixel fidelity, no Python needed, one-command install.
Give text-only models in DeepSeek Harness eyes, enabling image Q&A, long-screenshot OCR, UI restoration, and GUI visual tasks.
An armor-breaking plugin for DeepSeek: rewrites instruction priority, expands refusal-blocked phrases, persists across sessions, and shows a green active indicator.