DeepSeek Harness Plugins
DeepSeek Harness PluginManifest Valid

dsh-subagent-vision

Enables text-only DeepSeek sessions to read pasted or linked images via dedicated vision subagents.

VisionDeveloper ToolsBrowser & WebSecurity & PolicyRemote Execution
0GitHub Stars0ForksUpdated2026-09-16

Install

$ dsh plugin --profile web add dsh-subagent-vision

Plugin Overview & Capabilities

AI-assisted organization based on the public repository snapshot. The content must be grounded in source evidence and does not replace compatibility or security verification.

source-grounded
dsh-subagent-vision is a DeepSeek Harness bundle plugin that empowers text-only sessions (such as DeepSeek chat models) to process images seamlessly without switching models or opening secondary sessions. When an image is dropped or pasted in a text-only conversation, the browser component converts the draft attachments into secure temporary file paths at send time. The main agent then delegates vision tasks to an isolated one-shot subagent routed to a configured vision-capable model (like Qwen or experimental vision endpoints), returning textual analyses directly back into the primary chat.

Key Capabilities

  • Registers a specialized `subagent_vision` one-shot subagent tool pinned to a configurable vision-capable model route
  • Provides send-time image-to-path conversion in the web interface for text-only sessions, bypassing image admission rejections
  • Integrates a settings UI under '视觉处理模型' to discover, select, and declare image-capable models dynamically
  • Supports drag-and-drop and pasting images while retaining native composer thumbnails and undo behavior
  • Handles private temporary file storage with magic-byte validation (PNG, JPEG, GIF, WebP, HEIC/HEIF) up to 25 MB

Useful For

  • Analyzing screenshots, diagrams, and photos in DeepSeek chat sessions without manually toggling to multimodal models
  • Delegating OCR and document image summarization to high-performance vision models while maintaining conversation context in DeepSeek
  • Streamlining image uploads through paste-to-path automation in the DeepSeek Harness web interface

Who It Fits

  • DeepSeek Harness web users working frequently with images and screenshots
  • Developers looking to combine text-specialized models with dedicated vision subagent routing

Documented Limitations

  • Vision model routing is configured globally via settings or patch files and cannot be swapped per individual tool call
  • Subagent execution is one-shot; continuing multi-turn dialogue directly with the vision child requires standard subagent tools
  • The primary agent only receives text descriptions; the raw image pixels are processed exclusively within the delegated child log
  • Temporary files generated during paste conversion accumulate in the OS temp directory until cleared by system utilities

DSH Compatibility

Version-specific runtime evidence collected by DSH Plugin. A missing result means we have not tested that combination yet.

Not tested yetNo runtime compatibility tests have been published yet.

Security Signals

Objective signals discovered from package metadata and source inspection. These are not a guarantee that a plugin is safe.

No automated security signals have been published yet.

Source & Registry Notes

Public provenance, Registry classification, and the latest source check for this entry, kept separate from runtime verification.

Source repository
niuniuaba/dsh-subagent-vision
Registry source
GitHub · dsh-plugin topic
Registry classification
Plugin Bundle
Source checked
9c02042 · 2026-09-19

This project is independently indexed from public source information. DSH Plugin is not affiliated with DeepSeek or the plugin author. Always check the author repository before installation.

Repository Activity

GitHub stars
0GitHub stars
Forks
0
Open issues
0
Last commit
2026-09-16
Last release
No release detected
For maintainers

Maintaining this plugin?

This listing is generated from public repository data. If you maintain this project, you can review the information and share this listing with your users if you find it useful.

Add to README
Listed on DSHPlugin.app