Advertise image paste on text-only DeepSeek routes, describe attachments with a vision sidecar, and leave native vision models untouched.
Install
# from GitHub (first run asks for allowBuilds approval — follow the hint, retry)
dsh plugin --profile web add github:shinjiyu/dsh-plugin-multimodal
Any plugin you install runs third-party code with your own permissions — it can read your files, use your credentials, and reach the network, and tool approvals don’t sandbox it. GitHub-sourced plugins also run build scripts at install time — pnpm blocks those until you allow them, so an install can stop with ERR_PNPM_GIT_DEP_PREPARE_NOT_ALLOWED or ERR_PNPM_IGNORED_BUILDS; dsh prints the exact key to add under allowBuilds in your profile’s pnpm-workspace.yaml, and the install works on the next run. Allowing a build is a trust decision: only install sources you trust, and pin a commit (github:owner/repo#sha).
README
This plugin publishes its README in Chinese only.
DeepSeek Harness 官方线路是纯文本。Web 里贴图会被拒：当前模型不支持图片。
这个插件补的是贴图准入，不是视觉工具箱。
- 主模型本身收图（Claude / GPT / 自建视觉网关）→ 原样把图交给模型，不转文字
- 主模型是纯文本（官方 DeepSeek）→ GUI 先收下图，sidecar 转成文字再发给主模型
see_image给磁盘上的截图用
官方 PI adapter 已经会做第 1 步，不会做第 2 步：不支持就 UNSUPPORTED_CONTENT。Anionex 的 dsh-vision-toolkit 是另一条路：给 agent 一堆 vision_* 工具，要模型自己去调。本插件让粘贴不被拒。
对应需求：Discussions #588 · 介绍帖：Show and tell #1709
安装
dsh plugin --profile web add github:shinjiyu/dsh-plugin-multimodal
本地路径：
dsh plugin --profile web add D:\tempWorkspace\dsh-plugin-multimodal
然后重启 dsh web。旧会话的工具表不会更新。
GitHub topic：dsh-plugin
配置
环境变量（不要把 key 写进仓库）：
| 变量 | 作用 |
|---|---|
DSH_VISION_BASE_URL |
视觉接口，例如 https://api.example/v1 |
DSH_VISION_API_KEY |
该接口的 key |
DSH_VISION_MODEL |
必须是真能看图的模型，例如 glm-4.5v |
DSH_VISION_PROMPT |
可选。默认 OCR + 描述界面 |
没设 DSH_VISION_* 时回退 OPENAI_BASE_URL / OPENAI_API_KEY。GLM-5.2-FP8 这类文本模型不能当 sidecar。
也可以在 profile 的 cordis.patch.yml 里写：
- id: dsh-plugin-multimodal
name: dsh-plugin-multimodal
inject: [llm, tools, attachments, systemPrompt]
config:
model: glm-4.5v
apiKeyEnv: DSH_VISION_API_KEY
30 秒验收
- 主模型仍用官方纯文本（或 PocketCity 文本模型）
- 配好一个真能看图的 sidecar
- 在 Web 里粘贴一张报错截图，问「红字说了什么」
- 过关：不再弹「当前模型不支持图片」，主模型能引用图上的字
磁盘文件：让它调 see_image。
不要做
- 不要改
runtime/或官方 DeepSeek adapter - 不要把 sidecar 指到
deepseek-v4-flash/ 其它纯文本模型 - 不要把它宣传成「视觉工具箱」——那是 toolkit 的词
社区
微信群：deepseek harness 讨论群（2群）
扫码进群，用来分享和讨论插件。不是官方群。
微信群码大约 7 天过期（本张到 2026-08-23）。过期后把新码覆盖 docs/community/wechat-group-2.jpg 即可。
宣发稿（Discussions / V2EX / 群公告）在 docs/PROMO.md。
License
MIT
Links
More in this category
zhu1090093659/dsh-web-ui#packages/dsh-tool-describe-image★ 8121
A `describe_image` vision tool for text-only models: images (local path, URL, attachment) go to a configurable OpenAI-compatible vision endpoint and only the returned text enters the session.
liustack/modlens★ 4063
Vision bridge for text-only models: paste an image, get structured JSON evidence (OCR, layout, semantics).
ysr666/dsh-vision-router★ 1125
Free vision for text-only agents: built-in keyless vision chain plus pixel tools (Q&A, grounding, crop, pixel diff, colors, OCR, SVG trace, cutout, screenshots); paste an image to use it.
Anionex/dsh-vision-toolkit★ 884
Vision for text-only models: paste an image and the model switches to a Vision Toolkit variant for image Q&A, multi-image comparison, long-screenshot OCR, screenshot-to-UI reproduction, element grounding, and pixel diff. No API key by default — images are processed by the author-hosted free service, 100 per machine per day; configurable to your own provider.
dickpy/dsh-imagegen★ 93
AI image generation for the DSH Web GUI: text-to-image and image-to-image through a configurable OpenAI-compatible endpoint (gpt-image-2 / gpt-image-1 / dall-e-3), with an api_url/api_key settings card and a sidebar split-pane generation studio.
fandc520/dsh-comfyui★ 89
Drive a local or remote ComfyUI server from DeepSeek Harness: comfyui_run / comfyui_object_info / comfyui_workflow tools generate and edit images and videos, with a workflow library (graph extraction: per component / main flow / all), a load area with resolution auto-match, a live queue, SDXL and Wan 2.1 templates, a companion skill, and a same-origin media proxy.
Community comments
Comments are public GitHub Discussions. Loading them connects to GitHub and Giscus; a GitHub account is required to post.