@heeweelee/dsh-session-plugin
DeepSeek Harness plugin: recall previously sent messages in the input box with ↑/↓, and archive (delete) a workspace session with a right-click.
258 results
DeepSeek Harness plugin: recall previously sent messages in the input box with ↑/↓, and archive (delete) a workspace session with a right-click.
Declutter the DeepSeek Harness chat: hide agent execution-log rows (Think / context injection / tool calls), tighten message spacing and font sizes, wrap messages in rounded-corner bubbles, and inject an editable "reply concisely" instruction into the system prompt.
A local-only context badge with automatic first-five user-message capture, purpose summary, per-session notes and todos for DeepSeek Harness, plus an optional macOS foreground-window companion.
DSH web plugin: shows the current session's DeepSeek API cost as a small chip in the composer button row (conversation.input.right). Prices every assistant/message by its timestamp with the official DeepSeek pricing table, including the 2026-08-17 peak/off-peak policy.
DSH web plugin: 设置页 Token 消耗仪表盘 —— 总/输入/输出 token 指标卡、近7天每日 + 近24小时逐时趋势图、按模型/会话分布;宿主半扫描会话日志(assistant/message usage 事件)并提供 GET /usage/api/dashboard
Conversation position navigator (flat list): a right-edge tick rail with a hover panel listing every user message, click to jump (DeepSeek web-style).
DSH web client plugin: conversation-flow navigation rail on the left of the chat, plus a previous-message card docked at the top
A ChatGPT-style message navigator with hover previews for DeepSeek Harness conversations.
Codex-style user-message navigation for the current DeepSeek Harness conversation history
DSH 插件 v0.3.1:max-tokens 自动续写 + 429/限流自动恢复(agent/status 挂点、完整 user message、429 熔断)。嘻嘻,我一定要用 DeepSeek harness
用户消息超过12行自动折叠,保留产品默认气泡样式与操作按钮,支持 HMR 热重载
Inject Context for DeepSeek Harness (DSH): persist your own context entries and inject them as a standalone 'Remember:' row — right after your message — per turn or once per session (per-entry 'Every turn' switch). Simple/Advanced datasets, native Settings editor, standalone manager page.
DSH 插件:在聊天记录中自动将本地音视频/图片路径渲染为可播放的预览组件。When an assistant message or tool result contains a local media path, the path is replaced inline with a playable <audio>/<video>/<img> element backed by same-origin /api/media-preview/* route.
DSH Web 上下文窗口归因面板:通过 host token meter 按消息显示当前 surface 的 token 占用与摘要。Read-only per-message context-window attribution for DSH Web using the host token meter.
Codex 风格消息原地编辑(DeepSeek Harness / DSH 插件):hover 历史用户消息 → ✎ 编辑 → 同会话从编辑点截断并重新生成后续回复。
Session archive, cross-session read, message, wait, and copy-ID tools for DeepSeek Harness
Voice input for DeepSeek Harness: dictation chunked by pauses and voice messages, each with its own provider fallback chain (Deepgram, Groq, HuggingFace, local whisper.cpp, plus any OpenAI-compatible API of your own).
Auto-fold context/system messages in DSH conversation — hide the large system prompt / instructions by default, click to expand.
Session strata for the DeepSeek Harness Web GUI: the transcript's scrollbar becomes a to-scale, colour-coded map of the whole run — user messages emphasised, clickable anchors, auto-loading history, and a hover deck of every prompt in the session.
Cross-session project memory for DeepSeek Harness: seven-layer SQLite memory, first-turn snapshot injection, per-message keyword hits, memory_remember/search/project tools, automatic reflection with reflection-fold UI, and idle-triggered dream consolidati
DSH (DeepSeek Harness) web 消息增强插件:① 会话区左侧/右侧丝滑悬浮指令导航条(Codex 风格,可配置位置),刻度覆盖本地会话完整历史(含未加载部分,点击即加载更早并定位),滚动联动定位、悬停预览指令内容;② 输入框 ↑/↓ 键历史消息穿梭,像终端命令历史一样把已发送内容快速回填到输入框。
Daily activity brief for DSH: per-session turns/messages/tool-call stats across live sessions. Read-only.
Per-message delete for DeepSeek Harness — a trash icon beside the copy button on every chat message that removes that message from the derived model context (and the visible transcript) via the official surface-replace contract
Queue Consolidation for DeepSeek Harness: when the agent is busy and you queue several follow-up messages, the next turn runs with ONE consolidated formal prompt — the model rewrites the batch into a single prompt and executes once, instead of piecemeal one-message-per-turn rework.