DSH Plugin dsh-better-reasoning-effort: Reasoning Strength of Third-Party Models and Image Input Editing

dsh-better-reasoning-effort is an MIT-licensed plugin for DeepSeek Harness (DSH) designed to optimize the configuration experience for third-party models. The plugin allows users to directly edit model reasoning intensity (thinking levels) and image input support on the official model page, resolving the issue of scattered configuration settings. Its core features include a one-click "Auto-adapt" auto-adaptation via a built-in knowledge base and protocol reasoning, a "Stage" staging mechanism for unsaved model rows, and a quick reasoning intensity slider provided in the official menu. Users can declare image input capabilities through interface checkboxes, thereby enabling support for relevant attachments and tools. The plugin is suitable for scenarios requiring fine-grained maintenance of third-party model parameters, but it should be noted that the staging feature cannot express the state of being "intentionally unset."

Read More