dsh-experience-plugin

Daily use of DSH often surfaces a fragmented set of needs — model reasoning-level injection, CLI request mimicry, message folding, history auto-loading, global instructions — scattered across the interface with no unified configuration surface. dsh-experience-plugin was built to consolidate exactly these daily-optimization features into a single, well-organized settings area, complete with independent on/off toggles for every module.

The plugin delivers capabilities across three layers. At the model level, it injects reasoning tiers into custom API models and can disguise DSH requests as Codex, Claude Code, or Grok CLI payloads to match different backend conventions. At the conversation level, it adds a left-side message timeline rail with preview jumping, one-click collapse/expand of AI working steps, and automatic loading of earlier history when a session is opened. At the governance level, it offers a full-screen settings page, editor access to the ~/.dsh/AGENTS.md global instruction file, and per-module switches so you can adopt exactly the features you need.

It is tailored for developers who lean on DSH heavily for multi-model inference, frequently review intermediate AI reasoning steps, and want a single place to manage all daily configurations. If you are tired of tuning parameters manually with every model switch, losing track of key messages in long conversations, or lacking a unified entry point for global rules, this plugin streamlines the DSH experience considerably.