視覚ツール · DeepSeek Harness プラグイン — 40 ページ

Atlas Cloud skills and opt-in MCP tools for DeepSeek Harness.

DSH Web 输入框工具栏图片按钮：点击用系统文件选择器挑图片，不用拖拽。Toolbar button that opens the OS file picker to attach images to the DSH Web composer — no drag-and-drop needed.

附件自动降级：DSH 图片准入（单边 ≤2000px / ≤3.5MB / ≤4000万像素）超限图片不再拒绝，用 sharp 自动缩小/重编码后入库，用户上传手机原图不再踩线。

Codex-style attachment format expansion for the DeepSeek Harness Web GUI: PDF text-layer extraction (pymupdf4llm / pdfjs), Office text extraction, long-document spill + index cards, scanned-PDF OCR (tesseract.js), and browser-decodable images to PNG.

Codex-style attachment format expansion for the DeepSeek Harness Web GUI: PDF text-layer extraction (pymupdf4llm / pdfjs), Office text extraction, long-document spill + index cards, scanned-PDF OCR (tesseract.js), and browser-decodable images to PNG.

Bridges images into text for DeepSeek Harness: when the session's selected model cannot see images, a configured vision model describes them and the descriptions enter the durable session history as folded context rows — your message stays untouched.

DeepSeek Harness web plugin: a Model Capabilities settings page for declaring model input modalities, context window, output cap and image budgets without hand-editing settings.yaml.

Custom background wallpaper for the DeepSeek Harness chat page — upload a local image or paste an image link, tune opacity and brightness, scope to chat area or full screen.

Baidu cloud OCR (PaddleOCR-VL + Unlimited-OCR) for DeepSeek Harness Web: drag images/PDFs in, OCR to markdown, write results as local files.

DeepSeek 余额实时显示插件(含视觉模型): 在 dsh Web UI 输入框下方的统计条实时显示账户余额与本次对话的估算消耗, 内置 deepseek-v4-flash / deepseek-v4-pro / deepseek-v4-flash-vision-exp 官方峰谷定价

宝玉技能库（移植自 JimLiu/baoyu-skills）

Base64 文件转换：文件 ↔ base64/data URL（图片嵌入 HTML/JSON）

DeepSeek Harness background wallpaper plugin: upload an image and tune fit/scale/position/opacity/blur plus interface translucency, from a row at the bottom of Settings > General.

Custom background image for the DeepSeek Harness web UI (theme token overlay + settings card)

DSH 插件：给 DeepSeek Harness 换网页界面背景（系统预设/纯色/渐变/图片/视频）。侧栏「壁纸」按钮弹出右侧抽屉，滚轮+拖动的小图同时管缩放与定位，支持毛玻璃质感、本地媒体伺服、客户端即时生效并持久化到 settings；模型也可用 bg_apply 工具直接换。

DSH 自定义背景插件：为 DSH Web GUI 应用远程 URL 图片或本地图片作为背景，支持清晰度/压暗/模糊调节，设置面板保持默认外观。

背景皮肤：将 DSH 界面背景替换为随插件分发的默认图片（assets/ds.jpeg，安装即用），可选 imagePath 覆盖，缺图自动回退默认皮肤。

为 dsh-tool-vision 桥接导出的图片提供对话内联预览:用户粘贴的图片在气泡中显示缩略图,不改变桥接文本与模型行为

AI 浏览器操作：CDP 直连 Chrome，文本 snapshot+ref 主感知，截图走辅助视觉兜底

Browser automation for the DeepSeek Harness: CDP-driven tools that navigate, click, type, and read a real Chrome, with screenshots rendered inline in the conversation

DeepSeek Harness plugin + browser extension bridge: lets DSH agents read and control the user's browser, like Kimi WebBridge or Claude's browser extension. Node side: browser_* tools over a localhost WebSocket; browser side: a Manifest V3 extension.

Browser automation for DeepSeek Harness via bsk — navigate, snapshot, click, fill, and screenshot the user's real browser.

Read-only browser verification tools for the DeepSeek Harness web GUI: browser_open / browser_mock / browser_assert / browser_screenshot — verify a page (H5/desktop) in ≤4 tool calls with mock interception, DOM assertions, and screenshots that auto-project into the model context.

Browser tool for DeepSeek Harness that can see the page: drives a real Chrome over CDP with browser-use and reads it with deepseek-v4-flash-vision-exp, so canvas text, text inside images and rendered charts are readable. Schema-driven JSON extraction, Markdown output, and per-run token/cost accounting.