DeepSeek Harness Plugin Hub

发布与管理完整 Harness Profiles,发现适合你的插件。

探索

插件目录环境预设文档中心动态

社区

发布插件联系我们报告问题

相关链接

Plugin Hub GitHubDeepSeek Harness 官方项目系统状态隐私说明
© 2026 DeepSeek Harness Plugin HubPowered byPaxTech

独立、非官方社区项目,与 DeepSeek 官方无隶属、授权或背书关系。

Mimo Vision — DeepSeek Harness 插件(DSH Plugin)
← Plugins

mimo-vision

Mimo Vision

DeepSeek Harness (DSH) 原生插件:describe_image 工具,通过 ctx.fs / ctx.credentials 接口提供视觉桥接(图像 -> mimo-v2.5 -> 文本描述)

插件会安装到这里;不确定时保持 web。

npx -y @deepseek-ai/dsh plugin --profile web add mimo-vision@0.1.0
README兼容性版本

兼容性与来源证明

Mimo Vision 以 mimo-vision 发布,当前版本为 0.1.0。Plugin Hub 会校验它的 manifest,并保存精确安装来源,便于复现安装结果。

DSH 兼容范围
*
运行环境
any
发布来源
npm
Registry 更新时间
2026/9/20

版本

0.1.0stable
2026/8/16
查看其余 1 个版本收起版本
0.1.0-rc.5prerelease
2026/8/15

相关插件

正在加载相关插件…

最新版
0.1.0
DSH
*
HMR
重启进程
Tree shaking
未声明可安全裁剪
解包体积
54.2 kB
文件数
21
Surface
any
许可证
MIT
发布源
npm
GitHub
★ 1
周下载
64
最近提交
2026/8/16
查看源码 ↗项目主页 ↗
README Badge

点击下方 Badge 复制 Markdown,粘贴到 README 即可。

这是你的 Plugin?认领权益 · 优先安全扫描

验证 package.json 声明的 GitHub 仓库,即可管理这个公开页面。认领后,Hub 会优先安排当前版本的安全扫描,并在通过后公开展示结果。

认领这个 Plugin →
报告问题
DeepSeek Harness Plugin Hub
ProfilesPlugins分类动态文档登录管理 Profiles
ProfilesPlugins分类动态文档登录

README

mimo-vision · Native vision plugin for DSH

English | 中文

mimo-vision is a native plugin for DeepSeek Harness (DSH), package name mimo-vision. It registers a describe_image tool that sends an image to a mimo-v2.5-series multimodal model and returns the text description to the main model — a "vision bridge" built for main models (e.g. deepseek-v4-flash) that have no vision input of their own.

It is not a standalone process: it is a first-class citizen of DSH's "everything is a plugin" model. apply does exactly one thing — registers the capability as a first-class dsh tool. Dependencies, files, credentials, and subprocesses all go through dsh's defined capability seams; uninstalling cleans up cleanly.

Implementation paradigm: a direct landing of DSH's native plugin primitives

  • Registration is reversible by construction: apply(ctx) contains a single ctx.tools.register(defineTool(...)). register() returns a disposer; when the plugin fiber disposes, the tool is unregistered and its schema is automatically withdrawn from the system prompt. There is no leftover cleanup code — clean uninstall is a structural guarantee, not hand-written cleanup.
  • inject declares dependencies: export const inject = ['tools', 'fs', 'credentials'] follows the pure Cordis effect spec — activation only happens once the seams are in place. ctx.get('subprocess') in transcode.ts is an optional capability with a default fallback, used at execution time, not activation time. This is "declared dependencies", not "probed dependencies".
  • All capabilities go through dsh seams: file reads use ctx.fs.resolve/stat/readBytes + ctx.emit('fs/observed'); credentials use ctx.credentials.resolve(credentialRef(...)) with no hand-rolled parsing; subprocess (transcoding) uses ctx.subprocess. The one exception: transcode temp files are written via node:fs to the system temp directory (see Known Limitations below).
  • Uninstallable / composable: uninstall is pure — the disposer runs and everything is reclaimed: no disk writes, no timers, no long-lived connections to manually wind down.

Tool

ToolArgumentsDescription
describe_imagepath (required), question (optional)Describes an image file and returns text

Supported image formats:

  • Sent natively: PNG / JPEG / GIF / WebP / BMP (all verified decodable by the vision model)
  • Auto-transcoded: SVG / TIFF (.tif) / HEIC (.heif) / PSD / ICO / EXR / JP2 / JXL / AVIF — when ImageMagick is installed locally, these are transcoded to PNG before sending, downscaled to a ≤2048px long edge (saves tokens)
  • Any other extension, or a transcode-format when ImageMagick is not installed, returns an explicit error (never sent silently)

Usage example (tell the model): Use describe_image to describe D:\photos\cat.png, focusing on what breed of cat it is.

How it works

  1. Key resolution: ctx.credentials.resolve takes the first non-empty of OPENCODE_GO_API_KEY → OPENCODE_API_KEY (DSH credential layering: process env > ~/.dsh/.credentials.yaml > .env).
  2. Read the image: ctx.fs.resolve (relative paths resolve against the session workspace cwd) → ctx.fs.readBytes (20 MiB cap); non-native formats (SVG/TIFF/HEIC…) are transcoded to PNG via ImageMagick (through the ctx.subprocess seam) and downscaled to a 2048px long edge → base64.
  3. Routing: the free Zen route (mimo-v2.5-free) is tried first; on failure (non-2xx / timeout) it falls back to the paid Go route (mimo-v2.5) once per request; allowPaid: false disables the paid fallback.

Design decisions are documented in adr/0002-dsh-native-plugin.md (which supersedes the earlier MCP approach, ADR-0001).


Quick install (npm-installed DSH, recommended)

For npx @deepseek-ai/dsh web or a globally installed DSH. The repo ships prebuilt artifacts (lib/index.js) — no build toolchain required.

Prerequisite: Node ^22.19 || >=24, and DSH already able to start.

Step 0 · Install and activate (one command)

mimo-vision declares dsh.bundle (see the dsh field in package.json), so dsh plugin add reconciles it as a bundle layer of the profile — installing the package, mounting the layer, and activating the tool happen in one step:

# from the npm registry (recommended once this package is published)
dsh plugin --profile web add mimo-vision

# from GitHub (source install, prebuilt lib/)
dsh plugin --profile web add github:wulusai2333/mimo-vision

If the repo is not on its default branch, use github:wulusai2333/mimo-vision#<branch-or-tag>. This path relies on the dependency closure that DSH maintains in ~/.dsh/profiles/node_modules (symlinking all @deepseek-ai/* seams to the dsh install tree), so the plugin's runtime import "@deepseek-ai/dsh-tools" etc. resolves to the same instance DSH uses — singleton-safe, register()/inject semantics unchanged. GitHub-source installs require this repo to commit a prebuilt lib/ (kept in sync with src/): pnpm installs the source with lib/ included, no build runs, no allowBuilds needed.

Offline / manual mount from source (fallback path when pnpm is unavailable)

Manually copy the artifacts into the profile's plugin resolution root, then mount the patch in the profile's cordis.patch.yml (equivalent to automatic bundle mounting, but two manual steps):

# Windows PowerShell
$dst = "$env:USERPROFILE\.dsh\profiles\node_modules\mimo-vision"
New-Item -ItemType Directory -Path $dst -Force | Out-Null
Copy-Item package.json -Destination $dst -Force
Copy-Item cordis.patch.yml -Destination $dst -Force
Copy-Item lib -Destination $dst -Recurse -Force
# macOS / Linux
dst="$HOME/.dsh/profiles/node_modules/mimo-vision"
mkdir -p "$dst"
cp package.json "$dst/"
cp cordis.patch.yml "$dst/"
cp -r lib "$dst/"

Then edit ~/.dsh/profiles/web/cordis.patch.yml and add:

- insert:
    - id: tool-vision
      name: 'mimo-vision'
      config:
        allowPaid: true

Step 1 · Configure the key

Put an opencode key in ~/.dsh/.credentials.yaml (OPENCODE_GO_API_KEY preferred, OPENCODE_API_KEY as fallback):

OPENCODE_GO_API_KEY: sk-...

You can also put it in the environment of the process that starts DSH (OPENCODE_GO_API_KEY=... dsh web). Credential seam layering priority: process env > .credentials.yaml > .env.

Step 2 · Restart and verify

A restart is required on first integration (so the process imports the new package). After that, changes to this plugin's code or config such as allowPaid hot-reload without restart.

After restarting, verify: mimo-vision should appear in DSH's settings, with describe_image among the available tools; or just tell the model "use describe_image to describe some image" and try it.

Uninstall

dsh plugin remove takes the package name (the key in the profile's dependencies), not the install source:

dsh plugin --profile web remove mimo-vision

Passing github:wulusai2333/mimo-vision errors with ERR_PNPM_CANNOT_REMOVE_MISSING_DEPS ("no such dependency found") — the dependency is recorded by package name mimo-vision, so remove needs that key. The command removes both the dependency and the bundle layer.


Configuration (all optional, defaults provided)

FieldDefaultDescription
allowPaidtrueWhether to fall back to the paid route after the free route fails
freeBaseUrlhttps://opencode.ai/zen/v1Free route base URL
freeModelmimo-v2.5-freeFree route model
paidBaseUrlhttps://opencode.ai/zen/go/v1Paid route base URL
paidModelmimo-v2.5Paid route model

A minimal registration only sets allowPaid; everything else uses the defaults:

- insert:
    - id: tool-vision
      name: 'mimo-vision'
      config:
        allowPaid: false

Building from source / development (DSH monorepo)

To modify the plugin source, put it into the DSH source tree and go through the repo gates:

# 1. Copy this repo as packages/vision/tool-vision
cp -r /path/to/mimo-vision <dsh>/deepseek-harness/packages/vision/tool-vision

# 2. Install and verify (tsc typecheck + vitest unit tests + oxlint)
cd <dsh>/deepseek-harness
pnpm install
npx tsc -b packages/vision/tool-vision   # typecheck
npx vitest run packages/vision/tool-vision   # unit tests
npx oxlint packages/vision/tool-vision       # lint

# 3. Produce lib/index.js (prebuilt artifacts)
cd packages/vision/tool-vision
pnpm run build

pnpm run build runs tsc --build first (emitting the intermediate lib/types/*.js that tsdown consumes) and then bundles lib/index.js / lib/invariant.js. Steps 2/3 can also be run inside the package with pnpm run test / pnpm run typecheck / pnpm run build (the scripts are in package.json).

Note: the @deepseek-ai/dsh-* seams are versioned by the DSH dependency closure, not by this package. The peer dependencies are marked optional so a standalone npm install mimo-vision succeeds, but the plugin only runs inside DSH (where the closure provides those seams as a singleton). For development, either use the prebuilt artifacts from "Quick install" above, or run it from source inside the monorepo.

Failure semantics

Any failure (no key, unsupported format, both routes failing, file missing / not a regular file, over the limit, non-image response) is returned as a tool-level error: execute throws, the registry materializes isError, the process does not exit and the session does not break.

Known Limitations

  • Transcode temp files bypass the ctx.fs sandbox: when non-native formats (SVG/TIFF/HEIC…) are transcoded via ImageMagick, the source bytes and the resulting PNG go through node:fs to the system temp directory (os.tmpdir()), because ctx.subprocess needs real OS paths to feed the local magick, while ctx.fs's FsTarget may be an abstract/remote/sandboxed path. This bypasses the dsh-fs-sandbox file sandbox policy; the temp directory is deleted immediately after conversion. It only triggers when ImageMagick is installed locally and a transcode format is requested — native formats (PNG/JPEG/GIF/WebP/BMP) never take this path.
  • The prebuilt lib/ must stay in sync with src/: GitHub-source installs load the committed lib/index.js directly and never build at install time. After changing src/, rebuild and commit lib/, or the deployed version loads stale artifacts.

Security

  • The key is only read via ctx.credentials.resolve — never printed, never written to disk, never directory-scanned;
  • The key is only used in the request header Authorization: Bearer ...;
  • Images are read into memory via ctx.fs and sent straight as base64, never written to disk (non-native formats are routed through the system temp dir during transcoding and deleted immediately after).

License

MIT

相关插件

继续浏览 vision-media 分类下经过校验的插件。

Tool Describe Image@linxin666/dsh-tool-describe-image面向模型的 describe_image 工具,用于 dsh Web GUI:通过在兼容 OpenAI 的端点调用视觉语言模型,为文本模型提供图像理解能力,以描述一张图像(本地路径、http(s) URL 或附件引用)。可热插拔 —Modlens@liustack/modlens面向仅支持文本的 LLM 的插件视觉能力,由免费的 Antigravity CLI 提供支持Deepseek Ivideodeepseek-ivideoiPolloWork HyperFrames Video Studio,以及 27 个可编辑视频模板,以原生 DeepSeek Harness 对话视图呈现。Codexdsh-codexChatGPT OAuth、Codex 模型、搜索、read_image URL 支持,以及适用于 DeepSeek Harness 的 gpt-image-2 生成