dsh-model-search
Run the following command in DeepSeek Harness:
dsh plugin install GooDAnDReaDY/dsh-model-search
Paste the following prompt into your AI chat to install this plugin:
Run dsh plugin install GooDAnDReaDY/dsh-model-search in the DeepSeek Harness terminal to install; source code is available at https://github.com/GooDAnDReaDY/dsh-model-search
About this plugin
Running dozens of models across OpenAI, Anthropic, OpenRouter, Ollama, and DeepSeek in a single DeepSeek Harness workspace means the default dropdown turns into an endless scroll and a memory test for every provider/model-id. dsh-model-search injects a lightweight, non-invasive live search field right into the WebUI model picker so filtering becomes a keystroke instead of a scroll.
Key capabilities include multi-term search separated by spaces, commas, slashes, or semicolons; fuzzy subsequence matching (typing dsr1 hits deepseek-reasoner-1); an @provider prefix to lock results to a single provider; and a full ArrowUp, ArrowDown, Enter, and Escape keyboard navigation matrix. Under the hood the plugin only toggles DOM visibility, never touches the React virtual tree, keeps a selective MutationObserver at zero extra CPU cost during chat streaming, and renders entirely with DSH design tokens so dark and light themes stay seamless.
Built for heavy Harness users juggling multiple providers and a growing model catalog, and for keyboard-first automation workflows where mouse-driven scrolling is a bottleneck.
Use Cases
- Long model lists across providers make manual scrolling tedious
- Keyboard-driven filtering of a specific provider target model
- Partial recall of a model name needing subsequence matching to locate it
Best For
- Heavy Harness users connecting OpenAI, Ollama, OpenRouter, and other providers
- Automation developers who prefer keyboard-first workflows with minimal mouse use
- Team deployments managing dozens of models across multiple providers
Related Plugins
A unified suite combining hot runtime injection, task-aware thinking-mode routing, and a graded session protocol with red-team gates to sustain model diligence across long-horizon inference.
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.