starefinger/dsh-llm-qwen-local

Name: starefinger/dsh-llm-qwen-local
Author: starefinger

Plugin插件 Native原生 ⭐ 4 MIT Migration & Compatibility迁移与兼容 Vision & Media视觉与多媒体 Dev Workflow开发与代码工作流

面向 DeepSeek Harness(dsh)的 LLM 适配器插件:驱动由 OpenAI 兼容服务的本地部署 Qwen3.8-27B 模型。支持按模型多模态开关、完全可配置的推理档位、请求图像投影,以及中英双语 Web 设置页。

Project Overview项目介绍

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

A DSH adapter plugin that connects a locally deployed Qwen model (e.g. Qwen3.8-27B) served by vLLM via its OpenAI-compatible endpoint. Core capabilities: a per-model multimodal switch and fully configurable reasoning effort levels. Use it when running Qwen offline or behind a private vLLM server. Caveat: the multimodal declaration is unverified — true on a text-only endpoint fails mid-turn, false on a vision endpoint is silent.

DSH 适配器插件，连接本地 vLLM 部署的 Qwen 模型（如 Qwen3.8-27B）OpenAI 兼容端点。核心能力：每模型多模态开关、推理强度完全可配置。适用于需离线或私有部署 Qwen 的场景。注意：多模态声明不会验证，文本端开启视觉会中途失败。MIT，社区维护。

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-llm-qwen-local（starefinger/dsh-llm-qwen-local）
仓库：https://github.com/starefinger/dsh-llm-qwen-local
本站详情页：https://www.yhbd.top/plugins/starefinger-dsh-llm-qwen-local/
本站登记：类型 plugin · 归类 原生 DSH 插件 · 许可证 MIT · ⭐ 4 · 最近提交 2026-09-24 · 主语言 TypeScript

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 dsh-llm-qwen-local

把 starefinger/dsh-llm-qwen-local 加入你的 DSH 配置（web profile）即可启用。

READMEREADME

14.5 KB · 缓存于 2026-09-29 22:12:03 · 14.5 KB

dsh-llm-qwen-local

English | 简体中文

Qwen 本地 (vLLM) settings page

DeepSeek Harness LLM adapter plugin for a locally deployed Qwen model (e.g. Qwen3.8-27B) served by vLLM behind its OpenAI-compatible /v1/chat/completions endpoint.

v0.4.1 · exact compatibility target: DSH 0.1.2-rc.1 · MIT · community-maintained and not a DeepSeek or Qwen product.

✨ New in v0.4.1 — settings-page fixes; maxRequestImageBytes route cap removed

No more focus loss while typing a reasoning-effort id (or a model id) — list rows now key off a stable row identity instead of the id text, so typing no longer remounts the row.

Shorter, uniform field labels — long explanations moved into input placeholders; model-card columns are width-aligned.

maxRequestImageBytes (the per-request total image byte cap) is removed from config, schema, and the settings page. Every image is inlined once it fits its per-image budget (imageMaxPixels / imageMaxBytes are unchanged); an oversized request is refused by the backend LLM service against its own input limits. A leftover value in an existing settings.yaml is silently ignored — no migration needed.

"Discover models from endpoint" now probes with the key currently in the API Key field — a freshly typed key works without saving first.

The model list can be emptied — models no longer requires at least one entry: save an empty list and the route stays mounted but dormant (no selectable models), then re-populate via "discover models from endpoint" or a manual add.

✨ New in v0.4.0 — zero runtime @deepseek-ai dependencies

The published plugin no longer depends on any @deepseek-ai package at runtime — no schemastery, dsh-llm, dsh-settings, dsh-attachment, dsh-launch-environment, or cordis. Its only runtime dependencies are the MIT-licensed eventsource-parser and Node.js builtins.

Why: the plugin now reproduces every DSH seam it touches (adapter contract, failure snapshots, brand ids, API-key/attribution/launch-env helpers, the settings-namespace Config surface) as small local modules under src/harness/ plus a frozen, hand-owned configuration surface. It loads against the host's live services without importing the packages that define them — the same dependency posture as the dsh-llm-ollama reference implementation.

What does not change: external plugin behavior is identical — provider route qwen-local, settings namespace llm-qwen-local, the settings page, model discovery, and the wire dialect. The DSH compatibility target stays 0.1.2-rc.1. The @deepseek-ai packages remain dev-only type pins (their import type references are erased from the build), so existing installs keep working as-is.

Upgrading: drop-in — just dsh plugin --profile web add dsh-llm-qwen-local@0.4.1 (or your pinned snapshot tag). No configuration changes required.

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 starefinger/dsh-llm-qwen-local 的 README.md。在 GitHub 查看原文 →