視覚ツール · DeepSeek Harness プラグイン — 41 ページ

Vision bundle + plugin for DeepSeek Harness: the describe_image tool reads local images and asks any configured multimodal route, with zero core changes

让 DSH 直接操控你的浏览器，无需视觉能力：零 spawn 连接你已开启调试端口的 Chrome/Edge，截图、像素断言、执行页面 JS。Drive your already-running Chrome/Edge over CDP — screenshots, pixel assertions, page JS — no vision model, no per-use approval, zero child processes.

DSH 对话流增强：回合级思考 chip（实时时长/实时文字滚动）+ 工具调用聚合 chip + 共享活动抽屉 + 对话流卡片（步骤卡/总结卡，回合结束后才出现）+ 对话截图（assistant 消息相机按钮，无头浏览器渲染成图）+ download 下载工具（host 半身注册 wire 工具 + 进度路由，client 半身 keyed toolview 实时进度条/速度/ETA）——自 dsh-webui 移植。零 DSH 源码改动。

Renders chat Markdown image references () as a gallery under each closing assistant message

WebUI 聊天框拖拽/选择上传任意文件到会话工作区 uploads/ 目录（图片拖拽仍走内置流程）

Paste clipboard images as file PATHS only on configured text-only routes: for listed models the paste is intercepted, saved through dsh-drop-caret's session upload route, and the path lands in the composer — vision-capable models keep their native image pipeline untouched. v0.3.0: host half bridges textOnlyRoutes over HTTP (static client plugins get no module-pipeline config) and the session id comes from the conversation slot.

AI 女友助手（Digital-Sweet-Heart）：把 DeepSeek Harness 的 Web 界面改造成微信风格的虚拟女友聊天应用（好友列表 / 对话 / 人设设定 / AI 肖像 / 发图发视频 / 照片分析），支持自定义对话、视觉、文生图、文生视频四组 OpenAI 兼容 API

ChatGPT-powered coding, image generation and web search for DeepSeek Harness, with account and usage settings.

A DeepSeek Harness plugin that routes text to a base model and images to a configured vision model.

DeepSeek Harness tool plugin: image generation (gpt-image-2) via the Codex CLI

DeepSeek Harness Codex 社区接入 / Community Codex integration for DeepSeek Harness

DeepSeek Harness plugin: Codex image generation/edit/vision tools backed by the local codex-image-bridge skill.

Document and image analysis tools for agent harnesses, powered by a local OpenAI Codex CLI.

Codex-powered image understanding & generation for DeepSeek Harness: paste images into the Web GUI (converted to sha256 text refs via an injected paste bridge) and let the ChatGPT-Plus-backed Codex CLI identify / generate / edit images through a zero-dependency MCP server.

Use ChatGPT and Codex subscriptions in DeepSeek Harness with OAuth, quota, safe resets, web search, images, and Fast mode

Drive ComfyUI from DeepSeek Harness: generate and process images and videos through agent tools, with a workflow/asset/queue panel, in-chat results and a settings page. / 让 DeepSeek Harness 的 Agent 直接驱动 ComfyUI 生成与处理图像、视频，带工作流/资产/队列面板。

让 DSH Agent 通过自然语言全自动控制 ComfyUI:工作流列表/加载/UI→API 转换、任务提交与参数覆盖、进度监控(队列+WebSocket)、结果获取与下载、批量编排。

DSH web 插件：对话输入框 Markdown 编辑增强 —— 无序/有序列表自动续行、列表层级缩进阶梯（Tab 降级 / Shift+Tab 或前缀后 Backspace 升级，整棵子树联动，降级限父级下一级，顶层回归普通文本）、行内代码样式（保留字面反引号）、代码块围栏自动闭合与围栏内 Enter 保护。仅增强编辑态视觉与按键行为，发送文本保持字面 Markdown 逐字节保真。/ DSH web plugin: markdown editing aids for the composer — list continuation, the list level ladder (Tab sinks, Shift+Tab or Backspace-after-the-atom lifts, whole subtree riding, sinking capped at one level below the parent; a top-level lift unlists to plain text), inline-code styling (literal backticks kept), code-fence auto-closing with in-fence Enter protection. Edit-state visuals only; the submitted text stays literal markdown.

Image compression/conversion tools for DeepSeek Harness via api.compressio.cc MCP

DeepSeek Harness plugin: Linux X11 computer-use tools (screenshot, mouse, keyboard) via a Python XTest helper

DSH plugin: local macOS computer use — screenshot/mouse/keyboard/browser control + OCR fallback for dsh (DeepSeek Harness)

Independent cross-platform desktop computer-use plugins for DeepSeek Harness

Windows computer use for DeepSeek Harness built for a vision model: screenshot the real desktop, read the interface from pixels, and click what you see. Drives Win32 through koffi FFI with no native addon and no accessibility tree, and posts input to a window's own queue so the desktop focus is left alone.

Vision-only desktop automation plugin for DeepSeek Harness (fusion refactor)