dingminhua/dsh-subagent-default-model 预览 preview

dingminhua/dsh-subagent-default-model

Configurable default model for subagent delegations via settings.yaml, with single-model and multi-model round-robin/random strategies.

Project Overview项目介绍

This is a native bundle plugin built exclusively for DeepSeek Harness (DSH). It assigns default models to subagent requests that do not have an explicit model configuration specified, and leaves DSH core behavior unchanged for requests that already have a model set. It supports both single-model default routing and multi-model scheduling, allowing users to distribute subagent requests across multiple models from different providers via round-robin or random strategies. To install the published npm version, you can run the DSH plugin command: dsh plugin --profile desktop add dsh-subagent-default-model. You can also install it directly via npm with the command npm install dsh-subagent-default-model.

After installation, you can configure the plugin either through DSH's native web settings panel or by directly editing the local ~/.dsh/settings.yaml file. You can set a single fixed default model for all unconfigured subagents, or define a list of multiple models with custom allocation strategies and per-model reasoning effort settings. This plugin is ideal for DSH users who need to assign a uniform default model to subagents, or distribute load across multiple models to avoid rate limits or quota issues on a single model.

The plugin supports automatic failover to another model when a connection error occurs, which only triggers for specific error codes including rate limit, quota exhausted, server error, and timeout, and does not disrupt the main agent loop. It is released under the open source MIT license, which allows free use, modification, and distribution. Any install, update, or removal of the plugin requires a restart of the corresponding DSH process to take effect, but configuration changes do not need a restart. After uninstallation, leftover configuration can be manually deleted from the settings file.

这是一个专为 DeepSeek Harness (DSH) 开发的原生 bundle 插件,作用是为 DSH 中省略模型配置的子代理请求分配默认模型,显式指定模型的请求不会被插件修改,保持 DSH 核心原有行为不变。它支持单模型默认路由和多模型调度,可通过轮询或随机策略在多个不同服务商的模型之间分配子代理请求。

用户可以通过 DSH 内置的插件命令行工具快速安装已发布到 npm 的官方版本,安装完成后,既可在 DSH 的原生 Web 设置面板完成可视化配置,也支持直接编辑本地的 ~/.dsh/settings.yaml 配置文件。它适用于需要给子代理统一指定默认模型,或是需要多模型轮换分摊负载、应对单模型限流配额问题的 DSH 用户。

该插件支持连接失败自动切换模型功能,仅在命中限流、配额耗尽、服务端错误、超时等指定错误码时触发切换,主代理的工作循环不会受到任何影响。它采用 MIT 许可证开源免费使用,安装、更新或卸载插件后需要重启对应 DSH 进程才能生效,仅修改配置不需要重启。卸载后配置文件中的残留配置项可手动删除。

Pre-install check安装前体检Compatibility · Security兼容性 · 安全性 2 warnings2 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
  • No DSH plugin manifest detected - it may only carry the dsh-plugin topic, so the install method must be confirmed on the spot未检测到 DSH 插件清单:可能只是打了 dsh-plugin 话题,安装方式要现场确认
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 desktop add dsh-subagent-default-model

把 dingminhua/dsh-subagent-default-model 加入你的 DSH 配置(web profile)即可启用。

READMEREADME

dsh-subagent-default-model settings panel

dsh-subagent-default-model

为 DeepSeek Harness 子代理选择默认模型,并支持多模型轮换。

English · 安装 · 配置 · 更新日志 · 问题反馈

npm version npm downloads test status MIT license GitHub stars dshfind plugin

一个独立的 DeepSeek Harness (DSH) bundle 插件。它只处理省略了 agentOptions 的子代理请求;显式指定的模型始终优先,DSH 核心包保持原样。

亮点

三个核心功能

  1. 多模型分配策略(round-robin 顺序 / random 随机) —— 配置 ≥2 条路由后决定并行子代理如何分配:顺序轮换按列表依次取用、结果可预期(10 个子代理 / 2 条路由稳定 5/5);随机则无固定规律。分配发生在派发时,同一个子代理内部不会因策略中途换模型。
  2. 跨供应商故障转移 —— 子代理遇到连接类失败(限流/配额/服务端/超时/传输/空响应)或认证类失败(密钥无效/过期,AUTH/INVALID_CREDENTIAL)时,在本插件的 models 列表内换一条路由重试。候选来自整个列表,因此可以切到另一个供应商的模型。认证类也会切换(各家密钥独立时,一家密钥坏不该让任务死掉),但必打 warn 日志说明该供应商密钥可能无效,避免静默掩盖配置问题。耗尽即放行真实错误,主代理循环不受影响。
  3. 当前路由可见(你 + 子代理都能看到) —— 两条独立通路:
    • 界面行(给你):轨迹视图与对话视图显示「当前供应商/模型:provider/model」,故障转移切换时显示「已切换到」,会话恢复显示「继续使用」。
    • 提示词注入(给子代理):向子代理自己的上下文注入一行真实提示词,写明当前 provider 与 model,因此它能如实回答自己跑在哪个模型上;路由因故障转移变化时,这一行会在下一步跟着更新。默认开启,可在设置中关闭。

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 →

← 上一个 Prev dsh-jev-prune 下一个 Next dsh-commandcode-quota →