DeepSeek Harness PluginManifest Valid

dsh-design-qa

Adds an on-demand vision tool to DeepSeek Harness so tool-calling text models can inspect images for design-fidelity QA.

VisionUI & ProductivityTerminal & TUIDeveloper ToolsSkills & Workflows

44GitHub Stars3ForksUpdated2026-09-07

Source

Install

$ dsh plugin --profile web add dsh-design-qa

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-design-qa is an installable DeepSeek Harness plugin that exposes the deepseek_vision tool. It routes image inspection to a configured multimodal provider while keeping images out of the primary text model context unless the tool is called.

Key Capabilities

Provides deepseek_vision(image_path, question) for local image paths, HTTP(S) image URLs, and in-process attachment IDs.
Lets eligible tool-calling text models request multimodal image analysis through a separate configured vision route.
Includes configuration defaults for Alibaba Bailian and qwen3.8-max, while allowing another declared multimodal provider and model.
Offers an installer that can configure a route, apply or revert DSH host patches, and cold-restart DSH.
Includes a design-QA skill; the repository also documents benchmark fixtures and raw evaluation outputs.

Useful For

Assess whether an implementation matches a design mockup.
Let a text-only DSH coding model inspect pasted, local, or web-hosted images on demand.
Run evidence-oriented visual design review workflows with a separately configured vision model.
Evaluate replacement vision models using the repository's benchmark materials.

Who It Fits

DeepSeek Harness users building or reviewing UI implementations.
Developers using tool-calling text models in DSH.
Teams that need an image-analysis tool in a DSH design QA workflow.

Documented Limitations

The calling text model must support tool calling; otherwise it cannot invoke deepseek_vision.
The configured vision endpoint must actually support multimodal image input and be declared with input: [text, image].
Pasting images requires host modifications unless installed with --no-patches; without patches, paths, URLs, or attachment IDs can still be supplied.
Host patches persist on disk and must be explicitly reverted before uninstalling; DSH upgrades can overwrite them.
Attachment IDs are process-local and stop resolving after a restart.
The tool analyzes one image per call; multi-image comparison requires the caller to compose images first.
Private images requiring a login session may fail when passed as URLs.
The benchmark materials are documented as repository-only and not distributed with the package.

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.

Dependency Counts

package.json declares 0 runtime, 1 development, 5 peer, and 0 optional dependencies.

info

Dsh Bundle Declared

package.json declares DSH bundle metadata.

info

License Declared

GitHub reports the repository license as MIT.

info

Package Manifest Available

A root package.json was captured and can be inspected by the registry.

info

Source Available

Public GitHub source metadata is available for this registry snapshot.

info

Source & Registry Notes

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

Source repository: sunxin-ai/dsh-design-qa
Registry source: GitHub · dsh-plugin topic
Registry classification: Plugin
Source checked: 4c19310 · 2026-09-15

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: 44GitHub stars
Forks: 3
Open issues: 0
Last commit: 2026-09-07
Last release: 2026-08-20

Related DSH Plugins

Ranked by overlapping capabilities, use cases, plugin type, categories, and DSH profile.

modsearch

DeepSeek Harness web-search plugin that adds web search, X search, and focused page fetching with structured evidence.

View plugin →

dsh-cost-meter

A bilingual DeepSeek Harness cost-tracking plugin with session and daily costs, budgets, pricing management, balance lookup, and coding-plan quota displays.

View plugin →

dsh-vision-router

Vision routing and pixel-level visual tool suite for DeepSeek Harness agents with built-in free fallbacks.

View plugin →

pi2dsh

Compatibility bridge and host ABI enabling unmodified Pi plugins to run natively on DeepSeek Harness.

View plugin →