dsh-design-qa
Run the following command in DeepSeek Harness:
dsh plugin install sunxin-ai/dsh-design-qa
Paste the following prompt into your AI chat to install this plugin:
Execute dsh plugin install sunxin-ai/dsh-design-qa in DeepSeek Harness to install this plugin; the source code is located at https://github.com/sunxin-ai/dsh-design-qa.
About this plugin
While DeepSeek models excel at code generation, their inability to "see" images creates a bottleneck for automated design verification and fidelity checking. The dsh-design-qa plugin bridges this gap by giving text-based models a pair of "eyes." It operates via a tool-calling mechanism where the model autonomously decides to invoke a vision API when necessary, rather than flooding the main context window with image data. This approach is cost-effective and flexible, supporting any LLM that declares tool use.
Designed specifically for product design workflows, this plugin enables a "Design to Code to Verify" loop. It allows you to generate code from a design mockup and have the model judge whether the implementation matches the original intent. By leveraging a high-precision multimodal model (Qwen 3.8-max) as the vision engine, it ensures reliable detection of design discrepancies. It is an essential tool for developers and product managers looking to streamline the transition from design concepts to production-ready code.
Use Cases
- Generate code from design mockups and automatically check implementation fidelity
- Judge differences between design mockups and implementation code for auto-fixing
- Enable text-only models to recognize images for development assistance
Best For
- Developers building automated design review workflows
- Product managers needing to convert designs to code and verify fidelity
- Researchers using DeepSeek Harness to extend model capabilities
Related Plugins
A method pack that makes AI coding agents plan against your real baseline, prove completion with fresh evidence, and reduce reworks and unsafe changes.
Turns the DeepSeek Harness session into a captain that builds a durable sub-agent team, splits goals into dependency-aware tasks, and coordinates work via direct messages and a live Web UI.
Gives coding agents design judgment, letting Claude Code, Cursor, and 70+ agents generate and iterate high-quality UI, presentations, and graphics right from the terminal.
Run the Pi ecosystem's plugins on DeepSeek Harness, unmodified, via a compatibility layer that implements Pi's public extension ABI on DSH's native services.