시각 도구 · DeepSeek Harness 플러그인 — 8 페이지

DeepSeek Harness plugin: configurable vision model with vision_read_image tool, composer-bar vision-model selector, and automatic image-to-text conversion for text-only main models.

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.

Pure React attachment atoms for the dsh web UI: draft-image rail, message image gallery, and original-image lightbox (zero cordis)

股市行情皮肤+功能插件：从 Trading Terminal 皮肤演进而来的自选跑马灯、实时行情、自选列表与持仓盈亏面板，点击跑马灯弹出个股日K图（蜡烛/MA/成交量）（A股/港股/美股/指数/加密/外汇，腾讯+Binance+Yahoo+Frankfurter 聚合）

On-demand vision for text-only DSH sessions: images become markers, and a vision_describe tool sends only image + question to an OpenAI-compatible vision model

Drop or paste images, PDFs, office docs, zips, videos and audio into the DSH composer as workspace file paths instead of model attachments — lets a text-only model agent read them via vision tools.

dsh plugin: recognize attached images with the built-in Windows OCR engine (Windows.Media.Ocr) and send only the recognized text to the model. Text models never receive image bytes; vision passthrough is opt-in.

Codex-style Appshots for DSH Desktop: capture the frontmost macOS window by pressing both Command keys and attach it to the current chat.

Clean desktop shell for DeepSeek Harness (DSH) as a DSH plugin — wraps your web profile in a native window, tray-managed backend, offline auto-reconnect, zero visual changes. DSH 插件形态的纯净桌面壳：复用现有 web profile，托盘管理后端，零视觉改造。

Unified static-image generation router for DeepSeek Harness: one image_generate tool with standard parameters, config-driven OpenAI-compatible engine routing (OpenAI Images, Zhipu CogView, and any compatible endpoint) with health-aware fallback, durable workspace attachment results, per-session quota accounting, an in-conversation result card, and a Plugins settings panel that stores API keys as credential references.

AI-assisted Minecraft for DeepSeek Harness: drive an autonomous Mineflayer agent (survival, mining, chat, live map) with vision/control tools, plus a built-in launcher (Microsoft sign-in, version download, game launch).

Free image reading (vision) and image generation skills for DeepSeek Harness — Zhipu GLM-4V-Flash (Google Gemini optional) for reading, SiliconFlow Kolors for generation.

Local OCR fallback for DeepSeek Harness (Web): when the routed model cannot accept image input, a pasted image is saved locally and its text read by PP-OCRv5 + ONNX Runtime — fully offline, no vision model required. / DeepSeek Harness 本地 OCR 兜底插件（Web）：当接入的模型不支持图片输入时，自动把图片存到本地并用 PP-OCRv5 读出文字，完全离线。

Markdown campus resume workbench: the agent edits resumes and CSS templates under jobhunt/, and the user previews and exports from the sidebar.

Lightweight vision assist preset and CLI for DeepSeek Harness.

Renders images, video, audio, PDF and Office documents inline in the dsh web UI through a display_file tool, streaming bytes over a signed HTTP route with range support.

deepseek-aix right-panel system: a faithful port of the dsh-web-ui aionui-panel — the Explorer file tree with filename search, the multi-tab Preview (markdown/html/code/diff/csv/pdf/office/image/text/url), and the drag-a-file-into-the-composer dock, driven by a workspace-gated host filesystem service over /aionui-panel routes

Pure React attachment atoms for the dsh web UI: draft-image rail, message image gallery, and original-image lightbox (zero cordis)

Pure React attachment atoms for the dsh web UI: draft-image rail, message image gallery, and original-image lightbox (zero cordis)

Model-facing describe_image tool: asks a vision-language model at an OpenAI-compatible endpoint to describe an image file or URL

A DeepSeek Harness research workspace with archived PDF reading, optional OCR, human review, cited reports, and recoverable research tasks.

Image generation for DeepSeek Harness: a generate_image tool with pluggable providers — the FAL queue, any OpenAI-compatible images API, or a ChatGPT/Grok subscription with no API key at all. The picture is shown inline in the conversation; the model receives either a link (works with any chat model) or the image itself (needs dsh-vision-bridge or a vision-capable model).

DSH Web plugin that lets text-only models see images: paste images in the chat and send without file paths; the model discovers its own vision tools. Multimodal models pass through natively.

Movable, resizable Chromium desktop embedded in the DeepSeek Harness Web UI through noVNC.