Plugins

DeepSeek Harness (DSH) plugin: a "Thinking Language" selector under Settings → General that tells the model which language to reason in (chain of thought) via a system-prompt section. Zero extra calls, zero latency, 12 languages.

One-click collapse of every expanded section (Think rows, tool cards) with a live-count pill and a customizable hotkey.

DeepSeek Harness 插件：把思考状态行换成 287 条双语轮换词（Claude Code 158 + 鲸鱼系 13 + 中国本土雅 86 / 俗 30），一回合一个词，点击状态行切中英，词池可在前端编辑（双击 / Alt+E）。不改 DSH 源码。

Black-and-white theme for the DSH web client: UI tokens remapped to neutral ink greys, a Bayer-dithered signal field revealed by the pointer with click ripples, a blueprint-style headline that shows dashed selection boxes and dimension labels as the pointer comes near, neutral frosted composer and HUD glass, and a small school of pale-blue ASCII whales that drifts across the page. The new-session page and the chat page each have their own background, headline, whale and glass settings under Settings → Ink Whale.

Read .xd artboards, layers, text, styles, images, components, and prototype links. Render artboards and individual layers using your local Chrome or Edge browser. Export full-height HTML featuring selectable DOM text, transparent PNG graphics, component state transitions, hover/click interactions, and artboard navigation.

Site bookmark manager for DSH: a folder tree, multiple links per site, tags and filtering, with per-folder views.

Per-step thinking-effort pilot for DeepSeek Harness: simple tool rounds stay cheap, failures and heavy work lift one declared step, and agents can change their own depth through a set_effort tool without switching models.

P2P, end-to-end encrypted remote access to DeepSeek Harness, CodeX, Claude Code, Pi and other agent workspaces from web and mobile clients. Remote connections are coordinated through a hosted server (remote.gaolin.ink by default; you can point it at your own) and you sign in with an account there. It installs under the package name ds-harness-remote, the same as liguobao/ds-harness-remote, so install only one of them.

A Chinese-thinking plugin built for Chinese users: registers a system prompt section at order -900 so the agent reasons in Simplified Chinese.

Three neon palettes for DeepSeek Harness — cyan, pink and yellow — picked from a Theme colors row of three swatches under Appearance in Settings → General, each palette carrying its own light and dark token set; the built-in Appearance switch still owns dark and light.

Replaces the DSH startup screen with an ~18 s anime-style Link Start dive whose check, language, log-in and confirmation scenes show your own DSH status (profile, plugins, default model, workspaces, sessions, UI language). Click or press any key to skip; plays a gentle ~10 s version under reduced motion and lowers effect quality on slow machines. Unofficial fan work.

Shrinks the first history page of long DSH sessions to at most 100 messages, targeting the latest two turns, and fills the viewport in the background with the built-in load-earlier control when the trimmed transcript is shorter than the window.

Layout plugin for DSH: customize the structure, order, width and visibility of the left sidebar, right sidebar, conversation area and settings page. Drag-sort and resize the three columns, per-layer teardown of sidebar/conversation/right/settings, grid slot arrangement, draw-a-frame region relocation, entry consolidation into More menus, conditional linkage rules, multiple named layout schemes, and a fullscreen layout whiteboard with a four-layer structure tree. Config persists host-side; dangerous configs have multi-layer recovery entries.

The row that is running expands itself and folds back when it settles. A running turn keeps its whole procedure in one height-capped window that scrolls itself; a finished turn merges into the built-in finished header, leaving only the final answer. It works by driving the official disclosure rows rather than drawing its own panel, leaves rows you opened by hand alone, and restores the official settings on uninstall.

JetBrains Junie CLI as a one-shot DeepSeek Harness subagent provider over the Agent Client Protocol, with a settings page for model and thinking depth.

Reads local txt novels inside the chat dialog: chapters are detected automatically, the novel text is paginated in the assistant thinking block, a next-chapter button continues in the same reply, progress is remembered, and visible replies are fixed summaries.

Read-only diagnosis for a model provider that stopped working. Walks the connection chain one link at a time — host runtime and profile identity, the plugin row, whether the provider route is registered, whether it is configurable, the credential metadata the provider itself reports, the model catalogue, the provider's own diagnostic and the last recorded request — and labels each link observed, inferred or unavailable rather than guessing. Diagnoses the `workbuddy`, `workbuddy-ai` and `zcode-cli` routes from their plugins' public diagnostic commands, and adds a sidebar panel plus a copyable redacted report. The offline pass sends no model request, starts no process and reads no credential file; the connection check is a separate opt-in the panel describes before running it.

Paper reading in a DSH sidebar tab: the PDF keeps its original layout and fonts, a selection becomes a highlight, a note or a question in the current chat, and annotations plus paper-level notes are mirrored to Markdown in a per-paper folder that also holds the parsed text and the linked open-source code. Ships a `vibe_reading` tool and a reading skill so the agent can build that record itself.

Bilibili course notes for DSH: turns the videos and subject knowledge maps recorded by dsh-video-coursemap into same-name Markdown + Word notes with the video link on top, and delivers them in a new session through dsh-memory-evolve. Requires both plugins (with memory-evolve session orchestration turned on), Python with python-docx for the Word file (the DSH bundled runtime works), and Windows.

See the whole screen clearly: a 146-token overview map first, then 1:1 blocks up to 1400x900 so text is pixel-legible; a single sight_see tool with an on-screen eye toggle, measured against a 1600px host shrink limit. Windows only (it uses PowerShell); capture stays off until you click the eye.

Download queue with aria2 and optional Xunlei engines, magnet links, and a per-session history; aria2 is not bundled and must be installed separately, and the Xunlei engine needs its own SDK and API key. Built mainly for Windows.

Split long DSH conversations every 10 user messages by default, with todo handoff snapshots, source links and read-only archives; declares DSH 0.2.1-rc.5 or >=0.2.2-rc.1 <0.3.0-0 compatibility.

Per-workspace Fork library: a bookmark on each finished assistant turn stores that turn boundary as a full compressed snapshot that survives deletion of the source session; a "Fork 库" conversation tab lists a workspace's forks as a tree (a fork saved from a launched session nests under its parent) with name, turn count, last question and answer, note and source link, and "从这里启动新会话" seeds a new session from any fork, repeatable. Each fork then names itself from its own context: the snapshot is replayed to the model with a naming request appended, and that exchange is never written into the snapshot or a session.

Per-turn reasoning-effort routing that keeps the current provider/model: judges each user message (retry or correction, explicit intent, length, code blocks, errors, task scope) and sets reasoningEffort to a level the model actually declares, so Magpie and other OpenAI-compatible models work too. Ships a composer chip with an auto/manual switch and per-session decision history.