</>MCP Agents Market
Plugin

dsh-vision-toolkit

by Anionex874TypeScriptUpdated 2026-09-11

[dsh]为纯文本模型设计更强大的视觉工具箱：一行安装使用、粘贴图片直接识别、多张图片问答、截图到前端UI 还原等｜DeepSeek Harness-native integration for agent-vision-toolkit: image Q&A, long-screenshot OCR, UI restoration, grounding, pixel diff, Artifacts, and Web UI.

Claude Code

DSH Vision Toolkit is a DeepSeek Harness plugin that gives text-only models advanced vision capabilities including image Q&A, OCR, UI restoration, and visual grounding. It allows developers to paste screenshots directly into conversations, with the plugin automatically routing requests through a configured vision model while keeping deterministic image operations local. The toolkit combines ten specialized tools with battle-tested workflow playbooks that teach agents when to inspect, crop, trace, or compare images for complex visual tasks like converting sketches to working code or rebuilding UIs from screenshots.

Key Features

Paste images directly into DSH Web conversations with automatic vision-model routing and session history preservation
Ten specialized vision tools: image Q&A, grounding, element detection, cropping, SVG tracing, pixel diff, long-screenshot OCR, foreground extraction, color analysis, and HTML rendering
Task-aware vision that passes focus hints to multimodal models for answers relevant to the current step rather than generic descriptions
Bundled vision-skills playbook teaching agents complete workflows for UI restoration, long-screenshot reading, sketch-to-code, and GUI operation
One-command installation with automatic isolated Python runtime setup and fallback mirrors for reliability
Returns coordinates in original-image pixels that feed directly into subsequent cropping, tracing, or automation steps
Supports OpenAI Chat Completions and Anthropic Messages-compatible vision providers with configurable endpoints and credentials
Native DeepSeek Harness integration with Artifacts, workspace paths, Web preview, and lifecycle management

Use Cases

  • 01Convert UI screenshots or design mockups into working HTML/CSS pages with visual comparison verification
  • 02Extract complete chat histories and scrolling content from long screenshots with preserved structure and speaker attribution
  • 03Turn hand-drawn sketches or whiteboards into editable structured code (Mermaid, Graphviz) or functional interfaces
  • 04Automate GUI testing by locating controls from screenshots, performing actions, and verifying resulting states
  • 05Remove backgrounds from images or convert raster graphics to editable SVG vectors
  • 06Answer questions about pasted screenshots with context-aware analysis focused on specific user queries

Related Plugins

View more

andrej-karpathy-skills

A single CLAUDE.md file to improve Claude Code behavior, derived from Andrej Karpathy's observations on LLM coding pitfalls.

Plugin216.2k

claude-mem

Persistent Context Across Sessions for Every Agent – Captures everything your agent does during sessions, compresses it with AI, and injects relevant context back into future sessions. Works with Claude Code, OpenClaw, Codex, Gemini, Hermes, Copilot, OpenCode + More

Plugin95.1k

Understand-Anything

Graphs that teach > graphs that impress. Turn any code into an interactive knowledge graph you can explore, search, and ask questions about. Works with Claude Code, Codex, Cursor, Copilot, Gemini CLI, and more.

Plugin85k

rtk

CLI proxy that reduces LLM token consumption by 60-90% on common dev commands. Single Rust binary, zero dependencies

Plugin82.2k

dsh-vision-toolkit — FAQ

What is DSH Vision Toolkit?+

DSH Vision Toolkit is a DeepSeek Harness plugin that extends text-only AI models with vision capabilities through ten specialized image-processing tools and agent workflow playbooks. It enables image Q&A, UI restoration, OCR, visual grounding, and other complex visual tasks within DSH Web and Headless profiles.

How do I install the DSH Vision Toolkit plugin?+

Run 'dsh plugin --profile web add @anionex/dsh-vision-toolkit' for Web profiles, or target 'headless' or 'desktop' profiles as needed. After installation, restart your DSH profile and configure a vision model provider in Settings → Vision Toolkit with your API credentials.

Which AI clients does this plugin work with?+

This plugin works exclusively with DeepSeek Harness environments including DSH Web, DSH Headless profiles, and DSH Desktop. It requires DeepSeek Harness to function.

Do I need API keys or a vision model subscription?+

Yes, you must configure a vision model provider (OpenAI-compatible or Anthropic Messages endpoint) and store the API key as a DSH Credential. The plugin supports third-party aggregators like AIHubMix or local deployments of small multimodal models to reduce costs.

Is DSH Vision Toolkit free to use?+

The plugin itself is free and open-source under the MIT License. However, you will incur costs from your chosen vision model provider based on their pricing; using a locally deployed small multimodal model can minimize these expenses.

What prerequisites are needed to run this plugin?+

The plugin automatically manages its Python runtime, downloading a standalone Python 3.11+ if none is found on your system. For HTML screenshot rendering, Chrome, Chromium, or Edge must be installed; other tools work without a browser.

How do I install dsh-vision-toolkit?+

Open the source repository on GitHub and follow its README. dsh-vision-toolkit is a plugin — MCP Agents Market links you directly to the official repo.

Is dsh-vision-toolkit free?+

dsh-vision-toolkit is an open-source project hosted on GitHub. Check the repository for its license and any usage requirements.

Related searches

Developer Tools PluginsDesign PluginsProductivity & Workflow PluginsBest MCP ServersBest SkillsBest AgentsTop rated listingsClaude Code toolsBrowse all categories