dsh-model-presets

Picking the right model and reasoning-effort pair in DeepSeek Harness every single time interrupts flow, especially when you work across multiple API providers where the same model under two providers is genuinely two distinct configurations. dsh-model-presets lets you pin those go-to combinations as named presets so you stop re-selecting from scratch.

Open the model selector at the bottom of the composer, choose a combination, and tap the plus icon to save it. After that, switch with one click from the top of the list or call any of the first nine presets with Alt (Option) + 1-9. Presets can be renamed, reordered, and deleted, up to a cap of fifty entries. All data lives in local browser storage; nothing is uploaded, nothing syncs across devices, and presets survive a Harness restart. If a provider or model goes temporarily offline its preset dims and stops responding, then snaps back to normal once the service returns.

Built for DeepSeek Harness users who rotate between two or three fixed model-and-effort combos in a single workflow, have a strong preference for specific reasoning levels, or maintain parallel API providers. If you find yourself scrolling the model list and nudging parameters over and over, this plugin collapses that ritual into a single click or a keyboard shortcut.