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。

Llm Motomoto — DeepSeek Harness 插件(DSH Plugin)
← Plugins
L

dsh-llm-motomoto

Llm Motomoto

面向 MotoMoto 兼容 OpenAI 的端点的 DeepSeek Harness 套件:包含一个 pi-ai 提供商路由,以及“设置 - 插件”中的状态卡片。

插件会安装到这里;不确定时保持 web。

npx -y @deepseek-ai/dsh plugin --profile web add github:HandsYe/dsh-llm-motomoto#c377f72f2fec61e70f92036b7013a70b7b7dd6ca
README兼容性版本

兼容性与来源证明

Llm Motomoto 以 dsh-llm-motomoto 发布,当前版本为 0.3.0。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
web
发布来源
github
Registry 更新时间
2026/9/16

版本

0.3.0stable
2026/9/16

相关插件

正在加载相关插件…

最新版
0.3.0
DSH
*
HMR
重启进程
Tree shaking
未声明可安全裁剪
解包体积
未提供
文件数
未提供
Surface
web
许可证
MIT
发布源
github
GitHub
★ 0
周下载
0
最近提交
2026/9/17
查看源码 ↗
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

相关插件

继续浏览 models-usage 分类下经过校验的插件。

Usage@linxin666/dsh-usage用于 dsh Web GUI 的使用统计插件:检测各提供商的余额和编码计划配额,并提供实时令牌使用记录,同时在侧边栏条目中显示当前会话提供商今日的使用量Usage Stats@ychris12138/dsh-usage-statsdsh Web GUI 的令牌使用热力图、提供商余额和订阅配额Codex Connectdsh-codex-connect用于 DeepSeek Harness 的 ChatGPT OAuth 和 Codex 模型。Ui Usage Billing@kenz1117/dsh-ui-usage-billingDeepSeek Harness 的用量计费仪表板:侧边栏成本指标和完整的仪表板模态窗口,依据当前多提供商目录定价,并汇总会话日志中的实际用量。

README

dsh-llm-motomoto

A DeepSeek Harness bundle that adds MotoMoto to the model selector through the built-in llm-pi-ai adapter, plus a status card in Settings - Plugins modelled on the AgentRouter bundle.

The relay is Codex-only

Probing settled three facts, and the bundle is built around them:

ProbeResult
POST /v1/chat/completions (any User-Agent)Hangs — the path is not wired
POST /v1/responses + harness User-AgentRejected: bad_response_status_code / openai_error
POST /v1/responses + Codex CLI User-AgentStreams: response.created ...

So the route speaks api: openai-responses, and the runtime half fences globalThis.fetch to give relay requests the Codex client identity:

  • User-Agent is rewritten to the Codex CLI value (default codex_cli_rs/0.46.0 (Windows 10.0.26200; x86_64) WindowsTerminal) — the harness's own attribution header cannot be suppressed any other way: the pi-ai adapter strips configured headers colliding with it before appending its own.
  • originator: codex_cli_rs is added, matching the official client.
  • Request-body fields the Codex backend rejects are deleted from JSON bodies (default strip list: max_output_tokens — the official client never sends it, and the backend answers 400 Unsupported parameter otherwise). The list is the stripBodyParams setting: the next upstream complaint is a settings edit, not a release.
  • Requests to every other host pass through untouched; unloading the plugin restores the replaced fetch.

An upstream server_is_overloaded / 502 Upstream service temporarily unavailable is the relay's OpenAI pool being briefly busy — retry.

How the layers compose

dsh-settings resolves the llm-pi-ai section as mergeLayers(base, user): this bundle's patch is the base, the llm-pi-ai: section of ~/.dsh/settings.yaml merges over it — objects merge per key, arrays replace wholesale. A route you edit in the models settings page therefore wins field by field, while a fresh install still serves the default route before any user configuration exists.

The default route:

  • Base URL: https://motomoto.lol/v1 (OpenAI Responses API)
  • Model: gpt-5.6-terra (text input; low/medium/high reasoning efforts)
  • Credential reference: MOTOMOTO_API_KEY
  • Context window: 262,144

Route compat notes: the Responses protocol takes only supportsDeveloperRole, supportsStrictMode, supportsLongCacheRetention — completions-only keys (thinkingFormat, supportsReasoningEffort, chatTemplateKwargs, ...) on the same route crash the whole adapter at load.

Where to look after installing

  1. Model picker — the MotoMoto group.
  2. Settings - Plugins — the "MotoMoto" status card, reporting the live route (group, endpoint, credential reference, models).

Security notice

The API key is never stored in this package. Provide it as the MOTOMOTO_API_KEY credential.

Install from this source directory

git clone https://github.com/HandsYe/dsh-llm-motomoto
dsh plugin --profile desktop add "file:<克隆下来的仓库路径>"

Then fully restart the desktop app. Verify the load:

Select-String -Path "$env:APPDATA\DSH Desktop\logs\dsh-*.log" -Pattern 'llm-motomoto' | Select-Object -Last 5

Test

npm install
npm test

18 offline checks: patch shape, manifest, secret scan, the client bundle's registration and rendering, and the fence's identity rewrite, body strip, and unload behavior.

Uninstall

dsh plugin --profile desktop remove dsh-llm-motomoto