274 results
Model-facing describe_image + extract_text tools for the DeepSeek Harness web GUI: gives a text-only model reliable image understanding and OCR through an OpenAI-compatible vision endpoint, with critical-inspection prompts, auto-preprocessing, retries, and a persistent answer cache.
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.
DeepSeek Harness plugin: content-routing LLM adapter. Text-only requests go to DeepSeek; requests containing images go to a user-configurable OpenAI-compatible vision provider. Read_image works natively for text-only models.
DSH 视觉增强方案：识图技能（图片理解/OCR/文档解析）+ 幂等宿主补丁，让纯文本模型也能收发图片。
Persistent rembg GPU background-removal tool (remove_background) for DeepSeek Harness, in the dsh-vision-toolkit bundle form.
Bridges the four z.ai GLM Coding Plan MCP servers into DeepSeek Harness via the in-box @deepseek-ai/dsh-mcp-client: vision (GLM-4.6V stdio via npx @z_ai/mcp-server), web reader, web search prime, and zread repo MCP (streamable-http at open.bigmodel.cn)
Bridge Apple's on-device Vision framework (macOS) into DeepSeek Harness: OCR, image classification, face detection, and document layout as local dsh tools. No network, no API key, no daemon.
Vision bundle + plugin for DeepSeek Harness: the describe_image tool reads local images and asks any configured multimodal route, with zero core changes
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.
A vision bridge for text-only DeepSeek Harness models: registers an `analyze_image` tool that answers questions about images via ANY OpenAI-compatible vision/multimodal endpoint (SiliconFlow, DashScope, Zhipu, OpenRouter, Ollama, ...).
Workspace-owned durable process supervision, readiness checks, and a Runtime Center for DeepSeek Harness.
Zero-config local vision and realtime audio routing for DeepSeek Harness
DSH plugin that drives a local macOS camera gesture mouse helper.
Lightweight DSH upload bridge: images and common documents become workspace paths so a text-only DeepSeek model can read them with Qwen-MM-Plugins vision tools.
DSH Web 插件：调用智谱 GLM 视觉模型分析图片（analyze_image 工具 + 插件设置卡片）。
替换侧边栏左上角的品牌名与构建徽标文案（默认 DSH Local Build + commit hash），经 WebUI 插件配置页卡片实时配置。 | Replace the sidebar's top-left brand name and build-revision badge text (defaults to DSH Local Build + commit hash) via a live settings card in the Plugin Config page.
A canvas the agent draws on and then sees: eyes_render draws text/shapes in the Web GUI, stores the PNG locally, and hands the result back to the model. Windows-only: eyes_ocr reads text via the built-in Windows OCR engine (Windows.Media.Ocr).
PaddleOCR-VL-first visual model adapter for Codex and DeepSeek Harness
让文本模型（如 DeepSeek）也能正常粘贴并发送图片：图片落盘到工作区，消息转成仅模型可见的路径文本，由模型调用视觉/MCP 工具识别。
Transparent image guard + vision analysis for DeepSeek Harness: text-only models read pasted images without the 400 session deadlock.
Remote development for the DeepSeek Harness web GUI: connect a workspace to an SSH server, provision a .dsh-server there, run its files, terminals, conversations and agents on that machine — Work mode included, as the harness running there shown here
dsh bundle: subagent_vision — delegate image reading to a vision-capable model from a text-only session, plus paste-to-path so pasted images reach the subagent as file paths.
Agnes 全模态插件 for DeepSeek Harness：看图（agnes_vision）、生图/图生图（agnes_image）、视觉桥（带图消息可发送并自动交给 Agnes 视觉分析）。密钥经 DSH 凭据服务读取，不写入代码。
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。