Vision tools · DeepSeek Harness plugins — Page 55

Local vision 'eyes' for DeepSeek Harness (DSH): screen tool (capture screen or image -> local OpenAI-compatible VLM description) and ocr tool (Windows built-in OCR, zero model / GPU / cloud).

Model-facing generate_image tool for DeepSeek Harness (dsh): generates brand-new images with the Google Gemini image-generation service via the Antigravity CLI, saves them, and returns the paths. Hot-pluggable — install with `dsh plugin --profile web add dsh-tool-generate-image`.

DSH tool plugin: generate images through ToAPIs async GPT-Image-2 API (submit task, poll, download).

Standalone ocr tool plugin for DeepSeek Harness (dsh) — local image text recognition via the newbee-ocr (nbocr) engine, for models without vision input

Model-facing PDF tools for DeepSeek Harness: read metadata, extract text, and render pages for vision/OCR

Model-facing read_tiff tool for the dsh web GUI: decodes TIFF/TIF images (classic and BigTIFF, multi-page, LZW / Deflate / PackBits / CCITT / JPEG compression, bilevel, 8/16-bit and float) into a viewable PNG plus full header metadata, and — when a vision endpoint is configured — returns a text-only model's description of the image. Hot-pluggable via ~/.dsh/cordis.patch.yml + a profile node_modules link, no dsh source changes.

DeepSeek Harness 外置视觉模型插件：inspect_image 把本地图片或 http(s) 图片 URL 发给任意 OpenAI 兼容端点，视觉模型看图的文字回答直接带回对话；附 Web UI 设置栏。

Model-facing image_describe (识图) tool for DeepSeek Harness: describe a local image through the DashScope OpenAI-compatible API (qwen3.7-flash). Bring your own DASHSCOPE_API_KEY.

对话内联图片 DSH 插件 — 输入框/用户消息/LLM 回复中的本地图片路径直接渲染为图片(回环路由 + 强 token + 多根白名单 + 尺寸三档)。Inline local images in DSH web conversations.

Dependency-free dsh profile bundle bridging the native FTXUI DeepSeek Harness terminal surface

Fork of dsh-ui-appearance（毛玻璃外观插件，自审计自控版）— Appearance customization for DSH WebUI: theme palette, background image, frosted glass blur

DSH UI 自验驱动插件：UIA 驱动客户端 + 截图 + 视觉复核闭环（agent 工具 + 证据路由）

Windows UI Automation agent for dsh: list/snapshot/find/click/set_text/get_text/scroll/drag/swipe/screenshot + window state (topmost/minimize/maximize/restore/close). Permission-tiered (read-only / workspace-write / danger-full-access) with approval-based escalation. exe auto-releases from embedded assets.

Umi-OCR vision bridge for DeepSeek Harness: when the main model is text-only, OCR images with local Umi-OCR and let the original DeepSeek model answer.

DeepSeek 官网视觉主题：官网流体渐变背景 + 交互网格 + 粒子鱼 LOGO + 标题聚光灯 + 官网质感磨砂玻璃（双模式）

Unsloth for DeepSeek Harness - a pure-client tool plugin that connects the harness online model to a locally running Unsloth Desktop (Unsloth Studio) server for repetitive text and vision (OCR) labor.

Add an upload-image button beside the composer plus button. Picks local images and feeds them through the conversation draft-image drop path (DataTransfer + synthetic drop event), so describe-image rewrites the send into an image reference.

DeepSeek Harness plugin providing Add photos, Add files, and Uploaded files in chat UI with vision validation and session isolation

DSH user addons: file drop-upload, archived-session manager, model/token usage dashboard (auto-loads with the web profile)

DSH plugin: auto-tiles oversized chat images into labelled grid pieces (row/col metadata, overlap-aware, multi-image group isolation) so the DeepSeek vision model keeps fine detail instead of the ~800px downsample.

Replace DeepSeek Harness Web branding with Liang Wengu and switch logos during Beijing-time peak hours

A project-scoped multimodal video director plugin for DeepSeek Harness.

MiniMax H3 video generation plugin for DeepSeek Harness (easyframe single channel): text-to-video / first-frame image-to-video / first+last frame / last-frame extension / reference generation with async polling, queue & cancel tools, and an in-chat step-by-step Studio workbench

Bring text-to-video and image-to-video generation to DeepSeek Harness — DashScope Wanx, Volcengine/Doubao Seedance, Google Veo, OpenAI Sora & compatible relays.