qianshe/dsh-assistant-optimization

README 摘要

dsh-assistant-optimization A DSH plugin for the web profile: folds mis-rendered reasoning into collapsible blocks, renders mermaid inline, shows file-edit diff counts on collapsed tool rows, and rewrites composer drafts with the model you already selected. 中文文档 Features Thinking tag split — Some providers (OpenAI-compatible APIs without a dedicated reasoning content channel) dump chain-of-thought into the regular text output: DSH renders that as one block of message text. This plugin splits it at configurable markers so the reasoning collapses into DSH's native "Think" disclosure and only the answer stays visible. Mermaid diagrams — bash dsh plugin --profile web add github:qianshe/dsh-assistant-optimization dsh web 上次上下文：project 74 / instructions 237 (README.md) / summary 0 / asks 663 / results 420 Configuration Setting Default Description Markers [" "] Strings that split reasoning from body text. Multiple markers supported. Stored in localStorage under dsao:thinking-markers . How It Works Three tiers, each shadowing or extending a queried slot rather than replacing product UI. Message tier — reasoning fold Registered at conversation.chat.node key assistant-step , priority: -1 , sh…

分类