dsh-model-preset
Run the following command in DeepSeek Harness:
dsh plugin install kaixinbaba/dsh-model-preset
Paste the following prompt into your AI chat to install this plugin:
Install it in your DeepSeek Harness terminal by running dsh plugin install kaixinbaba/dsh-model-preset; the source lives at https://github.com/kaixinbaba/dsh-model-preset .
About this plugin
Switching between providers, models, and reasoning efforts in DSH Web usually means opening the side panel, scrolling, and confirming at each step — a small ritual you repeat every single time. dsh-model-preset pins your two or three go-to combinations as flat buttons in a full-width row above the composer card, compressing that ritual into one click.
The plugin does exactly one thing: add a shortcut layer to the user-facing controls. It injects no prompt text, does not alter tools, permissions, agent composition, or conversation structure, and sends no extra signals to the model. A visual settings editor lets you configure up to three buttons with labels, accent colors, providers, models, and reasoning-effort ids; changes take effect immediately and persist across restarts. A click routes through the DSH built-in model-selection path, so the next request in the current conversation uses the new combination while any in-flight request is left untouched.
It suits developers and researchers who hop between frontiers like GPT, Claude, and local models, or who toggle between high and low effort on the same model. If you want model switching to feel as light as flipping a tab, this is exactly what you need.
Use Cases
- One-click jump between GPT, Claude, and local models during a long session
- Toggle between high and low reasoning effort on the same model for A/B comparison
- Pin two or three team-shared combos so nobody has to scroll the model picker every time
Best For
- Developers who switch across multiple providers and models throughout the day
- Engineers benchmarking different reasoning-effort levels on the same model
- Researchers in DSH Web who tweak model parameters often and want fewer clicks to get there
Related Plugins
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.
An armor-breaking plugin for DeepSeek: rewrites instruction priority, expands refusal-blocked phrases, persists across sessions, and shows a green active indicator.