121103qwq/dsh-vision-sidecar

Name: 121103qwq/dsh-vision-sidecar
Author: 121103qwq

Plugin插件 Native原生 ⭐ 4 MIT Vision & Media视觉与多媒体 Sessions & Context会话与上下文

DeepSeek Harness 的免费托管视觉侧车，提供持久会话证据

Project Overview项目介绍

Edited by Zhiwei Agent (zhiwei)由知微Agent(zhiwei)编辑

This is a native DSH plugin built exclusively for DeepSeek Harness that adds visual perception capabilities to text-only models running inside DSH, without replacing the existing configured text reasoning model. Images get sent to a free or user-custom OpenAI-compatible vision API, and the resulting text description is stored directly in the durable DSH session for the text model to access. It ships with a default anonymous endpoint from LLM7.io that requires no account, API key, local VLM, or GPU to use out of the box.

The plugin is designed for DSH users who want to use text-only reasoning models to process images attached to their conversations. After installation via the DSH CLI command dsh plugin --profile <profile-name> add github:121103qwq/dsh-vision-sidecar#v0.1.4, it automatically resolves images from DSH’s verified attachment store, batches them for the configured vision endpoint, and saves the output to the persistent session. Future turns reuse the saved description, so you do not waste your VLM quota on reprocessing the same image multiple times. You can also configure it to use any OpenAI-compatible vision endpoint from providers like OVHcloud, OpenRouter, ModelScope, and Hugging Face.

This plugin is released under the open source MIT license, and has two core version requirements. It requires a DSH version of 0.1.0-rc.6 or newer within the 0.1.x release line, and Node.js version 22.19 or newer, or 24 or newer. The default anonymous LLM7.io endpoint comes with usage limits that currently stand at 500,000 tokens per day and 60 requests per hour, though these limits are subject to change by the provider. Users who add custom vision endpoints store their API keys via DSH’s built-in credentials service, and the plugin never stores or shares user-owned API keys itself.

这是一个专为DeepSeek Harness（DSH）打造的原生插件，用于给DSH中的纯文本大模型添加视觉感知能力，无需替换原有的推理模型。它的工作流程是将图片发送到免费或自定义的OpenAI兼容视觉API，再把生成的文本描述存入DSH会话，供文本推理模型读取处理。默认使用LLM7.io的匿名视觉端点，开箱即用，不需要注册账号、申请API密钥，也不需要本地部署VLM或占用GPU资源。

适合需要让纯文本推理模型处理图片的DSH用户使用。典型工作流是：安装插件后，DSH解析已验证附件存储中的图片，批量发送给配置好的视觉端点，生成描述后作为持久会话内容追加，后续对话会直接复用已有描述，不会重复消耗VLM配额。用户也可以自定义替换任意OpenAI兼容的视觉端点，支持对接OVHcloud、OpenRouter、ModelScope等多个平台的VLM服务。

本插件基于MIT许可开源，要求DSH版本为0.1.0-rc.6或更新的0.1.x版本，同时需要Node.js 22.19+或24+版本。默认匿名使用LLM7.io端点有配额限制，当前是每日50万令牌、每小时60请求，限制可能会发生变动。用户如果使用自定义VLM，需要自行保管API密钥，插件不会存储密钥内容，仅通过DSH的凭证服务读取。

View on GitHub在 GitHub 查看

Send to your DSH发给你的 DSH 安装

请帮我安装这个 DSH 插件。安装前先完成【兼容性检查 + 安全性检查】，检查通过再动手。

插件：dsh-vision-sidecar（121103qwq/dsh-vision-sidecar）
仓库：https://github.com/121103qwq/dsh-vision-sidecar
本站详情页：https://www.yhbd.top/plugins/121103qwq-dsh-vision-sidecar/
本站登记：类型 plugin · 归类 原生 DSH 插件 · 许可证 MIT · ⭐ 4 · 最近提交 2026-08-14 · 主语言 JavaScript

按下面顺序执行，每步先把结论告诉我，再进入下一步：

【1 兼容性检查】
① 我这边：DSH 版本、Node 版本、操作系统、当前 profile（web / desktop）。
② 读它的 README、package.json、插件 manifest，列出它要求的 DSH 版本 / Node 版本 / 操作系统 / 外部依赖 / 需要另外先装的运行时。
③ 逐条比对，结论只写「满足 / 不满足 / 未知」三种；不满足的给出可行替代方案。
④ 检查是否和我已装的插件冲突：命令名重复、skill / tool 重名、端口占用、重复注册的 MCP server。

【2 安全性检查】
① 仓库可信度：和上面「本站登记」是否一致；star / fork 数、创建时间、最近提交，是否归档或长期停更。
② 安装脚本：逐行看 package.json 的 preinstall / install / postinstall，以及 install.sh、setup.ps1 之类脚本。出现 curl|bash、下载后直接执行、混淆代码、访问与插件功能无关的域名，立刻停下来告诉我，不要继续装。
③ 依赖：列出新增依赖，标出无人维护、或与知名包拼写近似的可疑包（typosquatting）。
④ 权限与副作用：它会读写哪些目录、访问哪些域名、需要哪些 DSH 权限（filesystem / network / shell / clipboard 等），以及怎么卸载和回滚。
⑤ 如果它要求 sudo / 管理员权限，或权限明显超出功能所需，先停下来问我。

【3 安装】
上面两步没有「不满足」和「高危项」时才执行；用官方推荐方式安装，不要自行提权。

【4 汇报】
用表格输出：检查项 / 结论 / 依据 / 是否需要我决策。拿不准的一律写「未知」并说明要我怎么确认——不要猜，也不要替我决定。

Send this message to DSH in your current session: it verifies compatibility and security first (answering met / not met / unknown item by item) and only installs once everything checks out — it will stop and ask you if it finds a high-risk item. The box scrolls; the copy is the full prompt. CLI install commands may not be accurate across systems, so DSH is the safer route.把上面这条消息直接发给当前会话里的 DSH：它会先核对兼容性与安全性（逐条给「满足 / 不满足 / 未知」），确认没问题再安装，有高危项会停下来问你。框内可滚动，复制到的是完整提示词；安装命令不一定准确，发给 DSH 更稳。

Pre-install check安装前体检Compatibility · Security兼容性 · 安全性 1 warning1 项注意

Only 4 stars - very few users, little community feedback星标只有 4，几乎没人在用，遇到问题缺少社区反馈

DSH walks through these 9 checksDSH 会逐条核对这 9 项

Compatibility兼容性

DSH, Node, OS and profile requirementsDSH 版本 / Node 版本 / 操作系统 / profile 是否满足要求
External dependencies and runtimes (Electron / Python / Docker, ...)外部依赖与运行时（Electron / Python / Docker 等）是否齐备
Conflicts with installed plugins: command names, skill / tool names, ports, duplicate MCP registration与已装插件是否冲突：命令名、skill / tool 重名、端口占用、重复 MCP 注册

Security安全性

Repo matches the facts registered here; archived or abandoned?仓库是否与页面登记一致，是否归档或长期停更
Safety of preinstall / install / postinstall and install.sh / setup.ps1preinstall / install / postinstall 与 install.sh、setup.ps1 是否安全
curl|bash, download-then-execute, obfuscation, unrelated domains → stop immediatelycurl|bash、下载即执行、混淆代码、无关域名 → 立刻停止
Typosquatting or unmaintained packages among the new dependencies新增依赖里有没有 typosquatting 或无人维护的包
Requested permissions vs. what the feature actually needs申请了哪些权限、是否超出功能所需（filesystem / network / shell / clipboard）
Any sudo / admin requirement, plus uninstall and rollback是否要求 sudo / 管理员权限，以及卸载与回滚方式

Anything uncertain must be marked unknown with a note on how to confirm it. This site's signal screen is a static snapshot, not a security audit.拿不准的必须标「未知」并说明要我怎么确认。本站的信号筛查是静态快照，不能替代安全审计。

Or use CLI install (for developers)或使用命令行安装（适合开发者）

CLI Install命令行安装

dsh plugin --profile web add github:121103qwq/dsh-vision-sidecar#v0.1.4

把 121103qwq/dsh-vision-sidecar 加入你的 DSH 配置（web profile）即可启用。

READMEREADME

10.1 KB · 缓存于 2026-08-23 03:41:43 · 10.1 KB

dsh-vision-sidecar

中文说明

Give text-only models in DeepSeek Harness hosted visual perception without replacing the reasoning model. Images go to a free or custom OpenAI-compatible vision API; the exact description sent to the configured reasoning model is then committed to the DSH session and replayed as ordinary text.

The default is LLM7.io's anonymous default vision route. No local VLM, GPU, account, or vision API key is required for its documented anonymous allowance. No local VLM, GPU, or multi-gigabyte model download is required.

Why this plugin

No-key hosted vision default. On top of a working DSH text route, the default LLM7.io vision endpoint works without registration or a vision key; an LLM7 token is optional for higher limits.
Durable and replayable. VLM output is a real DSH session message, not a hidden request-time rewrite or process-only cache.
No image overhead for text. The vision provider is contacted only when an undescribed image exists.
Replaceable reasoning target. The sidecar forwards to targetProvider and targetModel; any DSH text route that does not depend on opaque provider replay state can be selected.
Fail-loud. Missing credentials, timeouts, rate limits, and provider failures remain typed errors. The plugin never silently forwards an image to a text-only model.
Build-free Git install. The repository ships native ESM JavaScript, so pnpm does not need permission to run a prepare script.

Requires DSH 0.1.0-rc.6 or newer within the 0.1.x line and Node.js 22.19+ or 24+.

Quick start: no-key hosted vision

Before starting, have a DSH Web profile that can already call its text model. The plugin does not require a particular reasoning provider or model; it forwards descriptions to the configured targetProvider and targetModel.

Make sure your DSH Web profile can already call its text model.
Install the plugin and start the Web profile. The default LLM7.io vision tier needs no vision account or key.

dsh plugin --profile web add github:121103qwq/dsh-vision-sidecar#v0.1.4
dsh --profile web

Showing the opening section of the README — the full document lives in the repository以上为 README 开头摘要，完整文档在仓库内 · View the full README on GitHub →在 GitHub 查看完整 README →

渲染基于仓库 121103qwq/dsh-vision-sidecar 的 README.md。在 GitHub 查看原文 →