媒体工具

ModLens 将用户粘贴或提供的图片发送到配置的视觉引擎，并为纯文本编码代理返回结构化的 OCR、版面和语义证据。

为文本模型智能体提供图像问答、长截图 OCR、界面还原和基于截图验证的 GUI 自动化工具与技能。

PicGo-Core 是一个支持命令行和 API 的图片上传引擎，可从文件路径或剪贴板上传图片并管理上传器配置。

为 DeepSeek Harness 文本代理提供图像路由、视觉分析及 OCR、裁剪、像素比较和截图等图像处理工具。

为文本模型提供图像问答、长截图 OCR、视觉定位、界面还原和像素对比等视觉任务工具。

A DeepSeek Harness plugin for finding, sending, learning, and managing meme image libraries locally.

为 DeepSeek Harness 提供图像理解能力，通过配置的视觉模型或本地 OCR 分析图片并辅助生成回答。

Provides text-to-image and image-to-image generation through OpenAI-compatible APIs, with history, gallery management, and prompt templates.

Provides text-only models with local image scanning, OCR, cropping, palette extraction, comparison, and document-to-image conversion, with optional external vision-model support.

Analyzes images and screenshots in DeepSeek Harness and generates images through configured model providers.

DirectorX is an AI media production plugin for DeepSeek Harness that supports image, video, and audio generation, storyboard canvases, local ffmpeg editing, and media quality checks.

An infinite canvas for DeepSeek Harness that uses the WeShop API to create and edit e-commerce images, virtual try-on results, and short videos.

Adds text-to-speech for DeepSeek Harness using Edge TTS or an optional local RVC voice service, with auto-read, playback controls, and audio downloads.

Connects DeepSeek Harness to a configurable external vision model for image recognition, screenshot analysis, and automatic return of results to the active conversation.

用于通过本地 MCP 工作流创作、调校、渲染、混音并审计原生 VOCALOID 项目的音乐制作插件。

Generates and displays images in the DSH chat window through compatible APIs or local CLIs, and analyzes local or remote images with available vision CLI tools.

一个通过节点连线画布组织文生图、文生视频、参考生视频和文本生成音频任务的媒体工具插件。

通过 CLI、MCP 或 DeepSeek Harness 将 DeepSeek 网页版视觉模型和 OCR 能力提供给 AI 助手，用于图片分析与文字提取。

将对话中的图片通过配置的视觉语言模型转换为文字，再交由 DeepSeek 生成回答。

Generates and edits images in chat through a configured third-party OpenAI Images-compatible API, with support for local reference images and saving generated results.

为 DeepSeek Harness 增加图像和视频分析能力，将媒体发送到配置的兼容视觉模型，并在 Web 侧边栏展示结果。

Provides image understanding, OCR, region inspection, and multi-image comparison through configurable web, API, browser, or local CLI vision channels.

Provides DSH agents with an image-understanding tool for OCR, UI analysis, debugging, and image description through configurable vision model providers.

通过兼容 OpenAI 的视觉模型 API 为文本模型提供图片描述、图像问答、OCR 和多图对比能力。

Uses Windows OCR to convert image attachments in dsh sessions into text for text-only models.

Registers Remotion skills for agents and provides practical guidance for creating programmatic videos with React.

This is a DeepSeek Harness plugin for TUI/Web that converts pasted images or image paths into text output for use by the assistant.

Provides on-demand visual analysis for image and document attachments in DSH minimal mode through a Bash workflow and a configured vision service.

Connects DeepSeek Harness to Timeline Studio to inspect, preview, edit local timeline projects, and render MP4 files.

通过 MCP 接入商汤视觉模型，为文本模型提供图像分析、OCR、界面转代码、图表理解和双图对比等工具。

通过兼容 OpenAI 接口的视觉语言模型，将本地文件、网络图片或附件引用转换为图片文字描述。

Adds speech-to-text input, hold-to-talk message sending, and spoken assistant replies to the DeepSeek Harness chat composer.

为 DeepSeek Harness 提供图像理解、截图审查和图像生成功能，并通过用户配置的外部模型服务进行处理。

Bridges pasted images to text-only models through session files and text markers while preserving native image sending for multimodal models.

为 DeepSeek Harness 提供图片理解、OCR 转写和文生图能力，并通过设置页面连接用户配置的服务提供商。

A draggable floating Bilibili video player for DSH with content search, channel browsing, autoplay, mini-player mode, and saved viewing history.

Provides agents with ffmpeg and ffprobe tools for media inspection, cutting, concatenation, transcoding, subtitle burning, extraction, and GIF creation.

A plugin that adds HyperFrames skills for creating videos from HTML and web content with animation, subtitles, voiceover, and rendering workflows.

Adds configurable semantic UI sound effects for task states, attention events, and button interactions, with sound packs, previews, persistent settings, and an API for other plugins.

用于在 DeepSeek Harness 中搜索、学习、管理并发送表情包图片的插件。

为智能体提供调用阿里云通义千问或火山方舟服务生成图片的 image_generate 工具，支持文生图及部分图生图参数。

为 DeepSeek Harness 提供图片描述、OCR、视觉问答、UI 布局分析和剪贴板截图分析等视觉工具。

Vision Mix adds image understanding, follow-up analysis, and image generation or editing through configured APIs to DeepSeek Harness.

通过可配置的视觉语言模型接口为 dsh 提供 AI 图像分析工具，并支持备用接口和设置页面。

为 DeepSeek Harness 提供可配置的外部视觉模型访问，以及图像分析、OCR、处理、截图和图像产物工具。

将粘贴的图片或工作区图像通过配置的视觉模型和 OCR 路由转换为文本，供 DeepSeek Harness 使用。

Provides agents with text-to-speech, audio transcription, and voice listing tools.

Registers a sidebar file viewer that plays supported local audio and video files inline with native controls.

A dsh web text-to-speech plugin that reads final assistant replies and approval prompts aloud with configurable voices, speed, autoplay, and notification sounds.

A draggable floating music player for DeepSeek Harness that searches and imports NetEase Cloud Music content and supports playback control through the interface or chat.

在 DSH WebUI 侧栏中浏览分类短视频，支持原生播放、直链解析、自动连播、手势导航和历史回放。

用于提取哔哩哔哩视频的字幕、评论、弹幕和关键帧，并生成视频分析摘要。

Adds image understanding and OCR, image generation and editing, and video generation to DeepSeek Harness with media outputs saved in the workspace.

Adds image understanding for text-only models by processing pasted images, local paths, or URLs through a configurable vision model backend.

Whale is a DeepSeek Harness plugin for AI video and motion-comic production, covering multi-stage planning, image and video generation, editing, rendering, and export.

Adds file and folder selection, drag-and-drop sending, and workspace attachment management to the DSH WebUI.

Manages an expression-image library and automatically selects and sends memes in assistant conversations based on emotion, preferences, and optional model analysis.

使用 macOS Vision 对用户提供的图片进行文字识别、表格结构提取和内容描述。

Adds a microphone control to the chat input and inserts locally recognized speech using the VocoType engine.

通过百度 AI Studio 的 ERNIE-Image-Turbo 模型根据文本提示生成图片，并支持预览、下载和插入会话。

为DSH Web实时合成打字机和机械键盘氛围音效，并提供音量、音色、节奏及静音设置。

一款提供可拖动浮动猫咪挂件、宠物图片与猫咪知识浏览、收藏及插件设置的轻量宠物体验插件。

通过浏览器麦克风将语音转换为文本，优先使用本地 FunASR ONNX 引擎，并支持浏览器回退和可选的 LLM 润色后发送。

通过模型可调用工具从 PixLuna 支持的图源获取图片，并支持 Pixiv 按作品 ID 获取与图源查询。

Converts image messages into descriptions generated by a vision model so text-only models can understand their content.

为 DeepSeek Harness 提供本地音乐搜索、播放、暂停、切歌、排队和播放状态查看功能。

Provides DeepSeek Harness with an image-understanding tool that accepts local paths or URLs and returns Chinese descriptions through configured multimodal models.

Displays user-uploaded images in the DSH chat interface and analyzes them with configurable cloud or local vision models.

A draggable floating NetEase Cloud Music player for the DeepSeek Harness Web interface with music browsing, synchronized lyrics, queue management, and account login.

Provides image upload handling, conversational image recognition, and dynamic bridging between text-only and natively multimodal models.

A DeepSeek Harness image-generation router that configures OpenAI-compatible providers and generates images through agent tools or a REST API.

Sends conversation images to a configured compatible vision model and supplies the resulting descriptions to text-only models.

Provides ambient soundtrack playback from AudioLib.ai, adapting local audio to the agent’s session state with controls for play, stop, and status.

dsh-vision is a DeepSeek Harness plugin that lets text-only models accept image-attached requests by turning images into text descriptions with vision or OCR before the model call.

This is a DeepSeek Harness chat companion plugin that lets each session define persona and voice, then automatically reads replies or speaks instantly via /speak and /cheer commands.

为智能体提供 read_image 工具，将工作区中的图片发送到配置的 OpenAI 兼容视觉接口并返回图像内容问答。

用于在 Windows 上检查本地视频或帧目录，并通过 RIFE 插帧生成 MP4 视频或 PNG 帧序列。

This plugin automatically sends chat images in DeepSeek Harness to a fixed vision model and injects the resulting observation back as hidden context for your main text model.

Adds image understanding, OCR, visual question answering, and reviewable persistent visual evidence to text-only DeepSeek Harness models.

一组为 DeepSeek Harness 提供图像理解、图像生成及统计显示优化功能的插件集合。

为 DeepSeek Harness 会话提供可配置的合成音效、提示音、静音控制和完成时的彩纸效果。

将拖入或粘贴的图片保存到会话工作区，并在纯文本模型对话中插入图片文件路径。

This is a DeepSeek Harness vision plugin that provides image analysis, OCR, grounding, detection, and color tools, with image input via local paths and paste-to-path image handling.

将图片发送至配置的视觉模型并转换为结构化空间文字，同时支持将编码描述或像素数据渲染为 SVG、PNG 或 HTML。

Ask questions about images through a selected multimodal vision model while viewing streamed answers and follow-up context in the interface.

通过粘贴、拖拽或选择文件上传图片，并调用可配置的视觉模型 API 进行图像描述、文字识别和问题回答。

Enables text-only DeepSeek Harness models to inspect pasted conversation images or image files through a configurable vision model and the read_image tool.

将本地或链接中的音频、视频配音翻译成多种语言，并通过远程服务返回配音结果。

Adds local full-duplex voice interaction to DeepSeek Harness with microphone input, speech recognition, agent-session bridging, and spoken playback.

A DSH Web plugin for browsing and playing Bilibili videos while pausing playback and notifying you when the agent needs attention.

将 Qwen-MM-Plugins 集成到 DeepSeek Harness，为文本模型提供图像、视频和文档理解，以及视觉、搜索和相关媒体工具。

通过 vision_read 工具将本地或远程图片发送给配置的视觉模型，并把图像分析结果以文本返回给 DeepSeek Harness。

通过 MCP 或 DeepSeek Harness 提供基于 Qwen 的图像理解、OCR、图像分析与文生图工具。

通过 Gemini 或 GLM 视觉 API 为文本模型提供本地图片描述、OCR 和视觉问答能力。

通过可配置的 OpenAI 兼容视觉模型读取图像文件并返回文字描述的 DeepSeek Harness 工具插件。

LocalLens uses macOS's built-in Vision OCR to extract text from screenshots, clipboard images, and image files for use by the agent.

通过 MCP 将本地或远程图像接入视觉模型，为文本模型提供 OCR、图像分析、界面检查、布局解析、截图重建和图像比较能力。

Adds image understanding to DeepSeek Harness by sending specified images to a configured external vision model and returning text descriptions.

Provides Sherpa-ONNX-based Chinese speech recognition, speech synthesis, and press-and-hold voice input for chat interfaces.

为 DeepSeek Harness 增加视觉能力，支持调用智谱视觉接口或使用本地千问-VL 2B 模型。

将用户粘贴的图片发送到可配置的兼容 OpenAI 接口进行视觉识别，并把识别文字交给纯文本模型理解。

A shared voice engine for consuming plugins that provides local Qwen3-TTS speech generation, browser audio playback, voice commands, session voice events, and optional scheduled greetings.

为 DeepSeek Harness 提供持久化视觉分析工具，可将本地图片或 data URI 交给可配置的多模态模型进行 OCR、截图、图表和通用图像分析。

Records microphone audio and transcribes it into the input field in real time using browser speech recognition, local offline models, or an OpenAI-compatible cloud ASR service.

为 DeepSeek Harness 提供语音输入、回复朗读、音频转写和文本转语音功能。

用于截屏录屏、将 Markdown 制作为图文卡片和 HTML、生成图片，并在人工审阅后分发到多个内容平台的内容创作插件。

Enables image-to-text understanding for text-only DSH models and optionally generates images through an external vision service.

Adds image understanding, OCR, region analysis, and multi-image comparison to text-only models through local or API-compatible vision backends.

This plugin adds image understanding and image generation tools to DeepSeek Harness, using configurable OpenAI-compatible or DashScope endpoints to process workspace images and return output paths.

Provides local image, video, and audio hosting for DSH with integrated preview, playback, and status controls in the WebUI.

This plugin generates a playable visual-novel web page in DSH from a script plus character, background, and audio/media assets, with in-client access for playback.

Adds voice input to the DSH conversation composer by converting speech to text with browser recognition or configurable OpenAI-compatible transcription APIs.

在 DSH Web 中选择对话范围并将其渲染为可预览、下载或复制的 PNG 长图。

Provides DeepSeek Harness tools for image recognition, image generation, and image transformation through configured external model APIs, with support for saving and displaying generated images.

将图片发送至可配置的视觉模型，并转换为包含空间坐标、元素关系和可选像素网格的结构化文字。

Adds browser-based voice input and assistant-reply read-aloud to DeepSeek Harness with configurable language, voice, rate, and automatic playback.

Analyzes specified or uploaded images with a configured local or OpenAI-compatible vision-language model and returns structured OCR, layout, semantic, visual, and uncertainty evidence.

Adds image question answering, target localization, and local tools for cropping, OCR, pixel comparison, color extraction, vector tracing, and foreground extraction to DeepSeek Harness.

Provides text agents with Windows screen or image descriptions through a configured vision endpoint and verbatim text extraction using built-in OCR.

通过 CDP 桥接桌面豆包，为 DeepSeek Harness 的纯文本模型识别聊天图片和本地图片。

Adds message playback, automatic announcements, and microphone dictation to DeepSeek Harness using configurable Alibaba DashScope, Volcengine, or browser system speech services.

Analyzes specified images through configured local OpenAI-compatible inference servers and generates descriptions or answers image-related questions.

Browse and play searchable YouTube Shorts and Bilibili vertical videos in DSH through a floating or edge-panel player.

为终端智能体提供语音转文字、文字转语音、回复朗读和音频卡片展示功能，并优先将音频存储在本地。

DSH Image Tools is a DeepSeek Harness media plugin for generating and editing images from chat using OpenAI-compatible and Gemini services, with configurable multi-service setup and saved outputs.

MindsEye adds image understanding, grounding, generation, and editing tools to DeepSeek Harness through configurable vision models with structured results.

一款基于 macOS Apple Vision 的本地视觉插件，可对图片、剪贴板、屏幕和应用窗口执行 OCR 及图像检测。

Generate GPT Image 2 images in DeepSeek Harness through a Codex subscription or API key, with progressive previews, durable attachments, preview, and download controls.

Enables text-only models to inspect uploaded images on demand through a view_image tool backed by configurable OpenAI-compatible vision APIs.

Converts speech to text through browser or OS dictation, applies constrained cleanup with the current session model, and inserts the result into the conversation composer.

Provides dsh with configurable image generation, video generation, and image recognition tools through grok2api or an LLM provider, with optional saving of generated media to the session workspace.

通过可配置的视觉语言模型分析图片文件，并为纯文本智能体返回图片描述或针对图片的问题答案。

This plugin adds third-party image generation to DeepSeek Harness, allowing chat-triggered text-to-image and image-edit workflows while providing a settings card for API model, key, and output configuration.

This plugin adds a `vision` tool for DeepSeek Harness text-only models, converting image file paths into text descriptions via configured multimodal models.

Captures a selected region or the full virtual desktop through browser hotkeys and saves PNG screenshots locally for users or agents.

Provides DeepSeek Harness agents with local text-to-speech, speech-to-text, audio playback, and human-confirmed voice-call interactions.

A Windows-focused vision toolkit for image understanding, OCR, cropping, screenshot comparison, color extraction, and allowlisted webpage capture.

Adds image recognition to text-based conversations by sending uploaded or local images to a configured OpenAI-compatible vision model with support for multi-turn follow-up queries.

通过阿里云百炼 qwen-vl 视觉模型，将粘贴或指定路径的图片转换为文字描述，供文本模型理解和处理。

Uses the MuApi service through one agent tool to generate image, video, audio, and 3D media and return result URLs.

Provides a describe_image tool that lets text-only models generate image descriptions from local image files or session attachments through a configured vision model.

为文本模型提供图像识别工具，支持读取本地图片或网络图片并通过可配置的视觉大模型返回文字描述。

为 DeepSeek Harness Web 输入框添加按住说话麦克风按钮，并在本机离线识别后将文字追加到草稿中。

Analyzes and transcribes images through configurable OpenAI-compatible vision endpoints, with support for image attachments and local archiving.

通过本地 LM Studio 视觉模型识别、描述和 OCR 处理本地图片或剪贴板截图，并提供智能体工具与设置面板。

为 DeepSeek Harness 的文本模型提供图像路由，通过分析附件或本地图像并返回视觉模型生成的文本描述来支持图像理解。

This plugin adds a text-based image handling path for DeepSeek Harness by converting input images into ASCII art text and sending it to a text-only LLM for content interpretation.

A DSH vision tool plugin that analyzes image files with a remote multimodal model and returns text descriptions.

通过视觉模型转录粘贴或本地图片，并为文本模型提供图像分析工具和视觉模型设置页面。

Submit images and questions through chat or a visual sidebar, then receive image analysis from the Zhipu GLM-4V API.

Accepts pasted or dropped chat images and uses a configurable vision-model endpoint to provide image understanding for text-only models.

将用户图片交给已配置的视觉模型分析，并把带像素坐标的文字描述注入对话，支持图片追问和设置管理。

DSH-Multimodal is a conversation-focused media input plugin that preprocesses image and media attachments into text through configurable model chains before sending them to the chat model.

通过 Codex Luna 将 DSH 中的图片附件转写为文字，并交给可配置的纯文本下游模型处理。

通过配置的多模态视觉 API 提供图像描述、OCR、元素定位、检测和裁剪等命令行工具。

Fetches Bilibili video metadata and subtitle timelines, with optional cover and representative frame extraction for agent-generated summaries with timestamp links.

A DeepSeek Harness vision plugin that converts images into structured JSON containing OCR, layout, and semantic information, with per-call vision engine selection.

Generates images from text prompts or edits supplied images through the local Codex CLI.

Adds inline previews for supported local video files in better-sidebar with playback controls, seeking, and downloading.

A ComfyUI integration that scans and imports workflows, preflights nodes and models, submits image or video generation jobs, and manages local outputs through contract-based tools.

Creates scores from lyrics and melodies and renders character singing as WAV audio through STUDIO NEUTRINO, with voicebank management and output listing tools.

通过图形界面或命令行从图片、视频和帧目录生成前景透明度遮罩，并输出合成与透明图像。

SideSight 为文本型编码智能体提供命令行媒体分析，可处理截图、图表、界面差异、视频和 OCR，并支持配置多模态服务或在 macOS 上执行本地 OCR。

将 SVG 路径和图层动画规格编译为可用于 Web、iOS 和 Android 的自包含 Lottie JSON 动画资源。

用于分析视频或音频链接并将支持站点的媒体流下载到本地文件。

MindSee is a DeepSeek Harness plugin that converts pasted screenshots or local images into text so DeepSeek can respond to questions about image content.

A DeepSeek Harness floating music player for QQ Music with search, playlist import, playback controls, and QR-code account login.

This plugin adds image-understanding and OCR capability to DeepSeek Harness text-only models by exposing tools such as describe_image and extract_text for generating descriptions and extracting text from images or documents.

为纯文本模型调用配置的视觉模型，将会话中的图片转换为文字描述并提供自动或手动识图能力。

This is a resident DSH vision plugin that auto-translates pasted images for chat, exposes image-recognition tools, and keeps session-scoped visual context for follow-up questions.

通过本地 stdio MCP 服务器调用外部视觉服务，为文本智能体提供图像理解，并支持实验性的图像生成与编辑。

用于规划图像、视频、音频和3D生成流程，选择模型并生成提示词、参数和分镜，也可选择性通过Atlas Cloud MCP执行工作流。

Enables image, OCR, and PDF understanding for text-only DSH models through a host patch and configurable vision services.

为文本模型提供图片分析能力，可读取本地或粘贴的图片并通过用户配置的兼容视觉端点回答问题。

Adds an analyze_image tool that reads local or pasted images and forwards them to a user-configured OpenAI-compatible vision endpoint for analysis.

Create, manage, render, and finalize long-form videos through AsimovBox’s hosted MCP service with asynchronous status polling.

Adds browser voice input, Xiaomi MiMo transcription and speech synthesis tools, configurable voices, and playable audio output to DeepSeek Harness.

Records speech from a selected microphone and appends transcriptions from browser, local Whisper, or Xunfei engines to the chat input.

A DSH web plugin for speech interaction in chat, enabling in-page voice playback and microphone-to-text input with configurable hotkeys and optional auto-send behavior.

A DeepSeek Harness plugin for role-play conversations, character-card creation and import, and image generation through supported interfaces.

This plugin maps chat images to local attachment paths and provides a vision tool backed by a configured API for models without native image support.

This plugin enables inline local image previews in DeepSeek Harness by using `img_serve` and `img` fence responses, and provides a standalone viewer for zooming and panning displayed images.

Live Talk is a DSH web plugin that displays and drives a virtual character in the conversation view while connecting media and speech capabilities for interactive dialogue.

Live Talk is a DSH web plugin that shows a virtual character in the conversation interface and animates its responses with speech-driven interactions.

Converts pasted or dragged images into text descriptions through a configured vision API and inserts or sends those descriptions in the chat input.

This plugin automatically runs MiniMax mmx vision on images shared in DSH chats and injects the results into the text-model context so image content is available to models during response generation.

Adds context-aware dictation to Composer by converting microphone input into editable text with optional local transcription, model polishing, and user-authorized automatic sending.

This is a DeepSeek Harness plugin that unlocks image attachment input and converts image blocks into text markers for use in image-handling flows on text-only upstream paths.

Embeds Bilibili, YouTube, Vimeo, or direct-link videos near the chat input and provides playback controls and keyboard shortcuts.

Reads and decodes TIFF/TIF images, converts pages to PNG, reports header and page metadata, and optionally uses a configured vision model to generate text descriptions.

A desktop chat plugin for generating and editing images through configurable OpenAI-compatible APIs, with inline display and attachment management.

Provides an agent with image generation, editing, compositing, and image-understanding capabilities through configurable external services, saving generated results to the workspace.

Converts images and selected document or video content into text for text-only models while processing image blocks before requests to prevent unsupported-image session failures.

DSH BG Tool is a DSH plugin that lets you perform image background removal from a conversation and save the transparent PNG result to a configurable local output directory.

This DSH web plugin provides local image OCR and image-understanding tools (such as ocr_image and view_image) and shows inline image previews in tool-result cards.

This plugin enables image sending for text-only models in DeepSeek Harness by saving images to session attachments and forwarding a model-visible path token for vision/MCP tool recognition.

This is a DeepSeek Harness multimodal chat plugin that routes image-based tasks to local or optional remote models and supports optional local realtime voice input with spoken responses.

Adds full-duplex voice interaction to DeepSeek Harness with microphone input, host-side speech recognition, streamed spoken replies, live captions, and voice interruption.

Recognize local images and generate new images through configurable multi-engine fallbacks, with OCR, layout details, and conversation UI integration.

This plugin adds `/music` full-screen scene and command-based controls in dsh-tui to scan local audio folders and play or control tracks through local media players.

为仅支持文本的 DeepSeek Harness 模型将图片消息转换为包含本地路径的文本占位，并通过视觉脚本生成图片描述。

Sends chat images to a configured vision model for text descriptions, provides those descriptions to the main model, and preserves the original images in chat history.

This plugin registers a video_understand tool for DSH to analyze Bilibili videos, BV IDs, or local video files and return structured understanding with answers.

This is a DeepSeek Harness image vision tool that uses a configurable OpenAI-compatible multimodal model to return normalized bounding boxes and optional annotated SVG output for spatial reasoning.

Extracts text from images with local OCR and can optionally use a configured vision model for additional image analysis.

A DSH web plugin that provides a NetEase Cloud Music player with search, playlists, lyrics, queue management, QR-code login, and conversational playback control.

This plugin provides voice input in DSH by recording speech from the chat interface and inserting the transcription into the input field, with local-first processing or optional cloud transcription provider support.

Adds a movable and resizable Bilibili video player to the DSH web interface that plays or pauses according to task status.

dsh-vision adds an image_analyze tool for Agent users to submit local images, URLs, or data URLs to a configurable vision model and receive text-based analysis results.

为纯文本智能体提供自动图片识别能力，支持通过云端兼容 API 或本地 Ollama 返回 OCR、布局、语义和视觉分析结果。

Adds image recognition to DeepSeek Harness conversations by routing images to native multimodal models or configured vision transcription services.

This is a DeepSeek Harness local music player plugin that scans local audio files and adds web chat controls with playlist and playback actions.

Provides local image OCR for DeepSeek Harness, extracting text for use by text-only agents.

This plugin for DeepSeek Harness Web adds speech input and response read-aloud capabilities, with manual/auto playback and configurable voice, speed, and pitch settings in the chat interface.

将用户图片附件交给配置的视觉模型生成文字描述，并回填到纯文本 DeepSeek 会话中。

Adds image understanding to DSH by using configured vision models to describe local images, image URLs, and follow-up visual details in chat.

This is a DSH system-level voice input plugin that records speech with a hotkey and pastes transcribed text at the current cursor, with optional recording storage and configurable settings.

通过可配置的 Meshy、混元 3D、Tripo3D 和 Rodin API，从文字或参考图生成并处理可导出的 3D 角色资产。

The plugin adds a comic sidebar entry in DeepSeek Harness to search and download JMComic manga, browse a local library, and read comics in an embedded reader with page navigation and resume support.

Reads DeepSeek Harness assistant replies aloud through the Fish Audio text-to-speech API with manual playback, auto-read, and voice settings.

An interactive Bilibili media downloader that uses the bilidown CLI to support part and quality selection, clipping, and local media processing.

This is a DSH Web GUI media utility that lets users configure event-linked sounds through a visual control panel, manage a local audio library, and optionally generate and add AI voices via an external TTS service.

Configure sounds for DSH interface events, manage a local audio library, and optionally generate voice clips with user-provided Alibaba Cloud Bailian credentials.

通过 Codex CLI 为智能体提供 image_gen 图片生成工具，并将生成的图片保存到会话工作区及可选地在 Web UI 中预览。

This plugin provides vision support for DeepSeek Harness by exposing a `vision_describe` tool that reads image input and returns a text description from an external vision model.

This plugin adds voice capabilities to DeepSeek Harness by enabling speech input via microphone or shortcut and playing AI replies with Edge-based text-to-speech.

This plugin renders Markdown image references into a gallery under assistant messages in the DSH web chat UI, including local-path and data URI handling with optional captions.

A voice-input plugin that records audio in the browser, transcribes it through a local SenseVoice service, and inserts the text into the DeepSeek Harness input field.

This plugin adds image-to-text support in DeepSeek Harness using local Umi-OCR for text-only workflows and exposes vision_* tools for OCR, text-region lookup, cropping, color analysis, and image comparison tasks.

The plugin adds a DeepSeek Harness `image_generate` tool for text-to-image generation, routing requests across configured OpenAI-compatible engines with fallback and returning attachment-based image results with session quota metadata.

This plugin adds voice interaction to DeepSeek Harness by turning microphone input into conversation text and speaking assistant replies through configurable STT/TTS engines, with optional event announcements and playback interruption support.

This is a browser-based DSH media plugin for local processing of video and images, including video metadata inspection, frame extraction, image compression, and GIF conversion.

通过 OCR 坐标识别、ASCII 布局图和可选的视觉模型语义描述，将图片分析结果以纯文本形式输出。

This plugin adds multimodal support to DeepSeek Harness, enabling image, audio, video, and document understanding, media transcription, and prompt-based image generation through Gemini services.

This plugin enables local image preview in DeepSeek Harness web chats by serving files from a local directory through an HTTP preview route and exposing a `preview_image` tool that returns an inline-capable URL for existing local images.

Uses Alibaba Cloud Bailian Qwen-VL to recognize uploaded or explicitly provided images and provide text descriptions to DeepSeek text models.

Uses a configured vision model to analyze chat attachments or local image files and return descriptions to text-only models.

This is a DSH image-understanding bridge plugin that automatically transcribes image message parts into text before model invocation so text-only models can continue with image-based context.

Provides DeepSeek Harness agents with image Q&A, OCR, grounding, image processing, pixel comparison, and HTML screenshot tools.

Generate images directly within DeepSeek Harness conversations by adding a chat-facing image tool, using your own provider API keys to produce and return images in the conversation flow.

This is a DeepSeek Harness plugin that provides a `generate_image` tool for model calls to generate images via an external service and return image URLs.

Analyze local images, image URLs, or data URLs with SiliconFlow vision models and return the results to the conversation.

Adds a describe_image tool that reads local images and returns text analysis from a specified multimodal model.

This plugin manages media attachment size handling in DeepSeek Harness agent-loop calls by applying budgets, optional compression, deduplication, and overflow handling, with an optional web interface to view media-guard reports.

This is a DSH plugin that displays successful `generate_image` results in the conversation as a thumbnail gallery, with options to view enlarged images and save them.

This plugin adds image-to-text support for chat and browser-screenshot workflows, turning image content into text for text-model workflows.

通过 NanoBananaPro API 为 DeepSeek Harness 提供图像和视频生成功能，并支持查询任务状态与 API 余额。

通过 Seedance 2 AI 公共 API 为 DeepSeek Harness 提供图像和视频生成及任务查询工具。

A macOS plugin that captures the current foreground window with a global shortcut and attaches the image to the active DeepSeek Harness Composer session.

This plugin adds a `local_vision` image analysis tool to DSH on web and provides a settings page to connect to a local or OpenAI-compatible multimodal model endpoint.

This plugin adds `grok_ask`, `grok_imagine_image`, and `grok_imagine_video` to DeepSeek Harness, invoking the local Grok Build CLI for text prompting and image/video generation and returning local output file paths.

Recognizes images pasted into DeepSeek Harness with Zhipu GLM-4V and injects the results into the conversation context.

Provides tools to inspect image dimensions and formats, resize images, and convert them to JPG, PNG, or WebP.

A DSH image plugin that reads image dimensions/format and performs resizing followed by image format conversion.

Uses user-configured language models to create short-video scripts, storyboards, captions, and platform-specific publishing copy, with optional video-model generation.

This is a DeepSeek Harness plugin that adds image understanding to text-only model flows by using a configurable vision model to convert chat images and Feishu/Lark document images into text descriptions.

This plugin adds local voice input to DeepSeek Harness Composer with a microphone button or shortcut, records audio, performs offline transcription, and inserts the result into the draft.

Analyzes local images through the DashScope vision API and provides agents with image descriptions or answers to focused image questions.

通过兼容 OpenAI 的视觉 API 为文本智能体提供图像理解、OCR 和多图比较能力。

Renders SVG logos into transparent animation frames and poster images, with optional MP4 synthesis.

将聊天中的图片通过本地 OCR 和可选的视觉语言模型转换为文本，并注入文本模型上下文以辅助理解。

Adds image understanding, text-to-image generation, image-to-image editing, and image handling in DeepSeek Harness conversations.

通过 Qwen API 为文本模型提供图像理解与 OCR、音频转写和文本生成图像工具。

提供基于 Sherpa-ONNX 的中文语音识别、语音合成和聊天界面按住说话功能。

This plugin adds a WeChat-style screenshot feature to the DSH web composer, allowing users to select and crop a screen region and send it as an image card in the chat input.

A music player plugin for local files and NetEase Cloud Music with Agent-controlled queues, playback, styling, and end-of-queue notifications.

Automatically saves images pasted into the chat composer to a configured directory and adds their absolute paths to messages for subsequent processing.

Provides tools to generate, edit, and understand images through a local Codex image bridge, with generated PNG files saved to a configured local directory.

This is a DSH Web GUI NetEase Music plugin that adds a sidebar launcher for searching and streaming songs, supports playing liked tracks after login, and provides a draggable mini/floating playback window.

This plugin adds image capabilities to DeepSeek Harness, enabling pasted-image input plus Codex-based image recognition, image generation, and image editing support in chat workflows.

Adds local Ollama-powered image understanding to DeepSeek Harness, including image description, OCR, element detection, comparison, and region queries.

A local OCR plugin that converts formula screenshots, textbook pages, and handwritten formula photos into LaTeX code.

Upload or paste images in the chat input and use a configured DashScope vision model or Windows offline OCR to describe them before adding the description to the outgoing message.

Provides realtime voice interaction for DeepSeek Harness by transcribing browser microphone input and speaking session responses through configured voice services.

Generates images through the asynchronous ToAPIs image API, uploads local reference images, and downloads generated results into the workspace.

Generates images through Gemini-compatible and OpenAI Images-compatible APIs and saves the results in the current workspace.

Provides hosted MCP tools for image compression, format conversion, lossless optimization, and metadata inspection.

将聊天中的图片转换为 OCR 文本和视觉模型描述，帮助纯文本模型理解截图、文档、照片和图表。

Analyze local images or image URLs with an OpenAI-compatible vision model and return a text response to a question.