dsh-model-modes: Inference Modes and Provider-Native Fast Mode for DeepSeek Harness Web Composer
dsh-model-modes is an MIT-licensed plugin (v0.1.3) maintained by DTSFO, designed specifically for the DeepSeek Harness web composer. Its core goal is to add an explicit `Fast ON/OFF` toggle while preserving the reasoning configuration in DSH's official model selector. The plugin addresses the inconsistency of Fast/Priority request parameters across different providers, and strictly ensures that enabling Fast mode does not silently switch models, downgrade reasoning levels, or change providers. It injects provider-native fast parameters at the final `llm-pi-ai` boundary via `AsyncLocalStorage`, supporting `Max + Fast` and `Ul
Read Moredsh-reasoning-slider: Switching reasoning gears within the DSH model selector
`qjcnmd/dsh-reasoning-slider` is an MIT-licensed DSH plugin designed to address the cumbersome process of configuring reasoning levels when frequently switching models in DeepSeek Harness. This plugin embeds the reasoning level slider directly into the model selector overlay, allowing users to intuitively view supported levels upon selecting a model and switch them in real-time by dragging the slider, with changes taking effect upon release. Additionally, the plugin supports keyboard arrow keys and mouse wheel operations, automatically retaining the current level when switching models. If the target model does not support the current level, it automatically reverts to the default level, significantly reducing operational overhead. Before installation, ensure that `pnpm` and `dsh` are available in the environment, then add the plugin via the `dsh plugin` command from npm or GitHub, and restart the `dsh web`
Read MoreAutomatically adjust the inference gears of DeepSeek Harness in rounds with dsh-tool-turbo
dsh-tool-turbo is a DeepSeek Harness host plugin maintained by Electricitysheep. It accesses the agent/request waterfall and automatically selects reasoning_effort based on recent tool calls: simple read and write operations will be downgraded to low, while heavy tasks will be considered for upgrading the level, and the task itself does not need to be modified. The directory installation command is `dsh plugin add github:Electricitysheep/dsh-tool-turbo`. This article is cross-verified based on the directory page, GitHub README/source code and the official llm-deepseek adapter.
Read More