dsh-multimodal-bridge
DeepSeek Harness plugin bundle: qwen_vision (Qwen-VL image understanding) and qwen_generate (Qwen-Image text-to-image and image editing) tools for text-only models
33 results
DeepSeek Harness plugin bundle: qwen_vision (Qwen-VL image understanding) and qwen_generate (Qwen-Image text-to-image and image editing) tools for text-only models
Share DSH Q&As or selected conversation groups as PNG or Markdown.
Session-aware Pencil integration for DeepSeek Harness with official MCP tools and an on-demand browser canvas.
DSKIN - cartoon pixel kittens for DeepSeek Harness (DSH): 1-4 random kittens spawn each session, they play together, you can pet them (hearts) and drag them around. Original UI untouched.
Eyes for text-only DeepSeek on DeepSeek Harness: image understanding via Doubao Web by default (zero cost, no API key), Antigravity IDE quota (flash/pro), Gemini API, or Cockpit proxy — model-invokable vision tool, wrapper adapters, evidence memory, and a bilingual client panel.
Persistent DeepSeek-inspired whale-dive and reactive-water animation for the Harness Web turn status.
Native Cordis WeShop canvas, tools, skills, and Web UI for DeepSeek Harness
DSH 视频创作技能插件:注册 Remotion 官方移植技能(React 编程式视频:动画/音频/字幕/3D/图表/字体,38 个规则文件),安装即用。
Eight DeepSeek Harness plugins: persona, language guard, vision fallback, python workdir guard, windows encoding guard, cross-agent memory, image generation, and skill shell injection.
DSH plugin: keep text-only models (deepseek-v4-flash / deepseek-v4-pro) as the session default, and automatically route requests that carry image content to a configured vision-capable model (deepseek-v4-flash-vision-exp) 鈥?no manual model switching (periscope).
趴趴鲸鱼娘 —— 一只趴在 DeepSeek Harness 对话框上的二次元鲸鱼娘桌宠:手绘 SVG、灵动动态、零外部依赖。
Paste images into DeepSeek Harness chat and have them read by a free local backend (macOS Vision / Tesseract) before the text-only DeepSeek model answers
DSH 产品宣传视频插件 —— 为 DSH Web UI 提供右侧工作区,基于 video-shotcraft skill 一键制作电影感产品宣传片
Shots tab for DeepSeek Harness: a video-player view (live + YouTube-style scrubber) over a browser daemon's screenshot feed in <workspace>/shots/
Image translation for non-multimodal models via GLM-4V-Flash: intercepts images, generates descriptions, injects as text.
Codex-style file search INSIDE the built-in dsh-better-sidebar Explorer tab (filter box under the workspace folder name), plus per-file-type icons/badges in the explorer tree and editor tabs
Configurable anchored popover and settings animations for DeepSeek Harness.
让 dsh 用独立的 OpenAI 兼容视觉模型读图:主对话历史只保留纯文本描述,图片字节不进上下文,纯文本模型也能读图。
常驻视觉服务:直连视觉模型(默认 opencode-go/minimax-m3,回退 zai-coding-cn/glm-4.6v)。describe_image / subagent_vision 工具 + 粘贴图片自动转译(llm/stream 钩子)+ 输入框视觉状态小胶囊与详情页(活动日志:指令/思考过程/输出)。零子代理、零 agent 上下文开销,按会话记忆窗支持视觉追问与验收。
为 DSH 设置侧栏的第三方插件提供统一的语义化图标与本地样式配置。
Zhipu BigModel capabilities for the DeepSeek Harness in one plugin: web_search_prime search provider, webReader fetch provider (server-side rendered), and GLM-4.6V vision (vision_analyze tool + pasted-image pre-step hook), all behind one ZAI_API_KEY
视觉层级构建参考
Visual plan mode for DeepSeek Harness: structured plan.json + plan.md, an editable React Flow canvas, comments, plan diff, versioned revisions, and reliable write-back to the agent.
Configure model input capabilities and reasoning efforts for DeepSeek Harness