DeepSeek Harness Plugin Hub

Publish and manage complete Harness Profiles. Discover Plugins for your next setup.

Explore

PluginsPresetsDocsNews

Community

Publish a pluginContactReport an issue

Resources

Plugin Hub on GitHubDeepSeek HarnessSystem statusPrivacy notice
© 2026 DeepSeek Harness Plugin HubPowered byPaxTech

Independent and unofficial. Not affiliated with, authorized by, or endorsed by DeepSeek.

Think Level — DSH Plugin for DeepSeek Harness
DeepSeek Harness Plugin Hub
ProfilesPluginsCategoriesNewsDocsSign inManage Profiles
ProfilesPluginsCategoriesNewsDocsSign in
← Plugins
T

@local/think-level

Think Level

Session reasoning levels: Set a reasoning level independently for each session page (Off/Low/Medium/High/Highest/Ultra High); sub-agents inherit the level of their session, with an additional global default level.

The plugin will be installed here. Keep web if you are unsure.

npx -y @deepseek-ai/dsh plugin --profile web add github:373831615/dsh-think-level#69bd87f994da13a5953a9a1666da305b12a1f05b
READMECompatibilityVersions

Compatibility and provenance

Think Level is published as @local/think-level and currently resolves to version 1.0.0. The Hub verifies its manifest and preserves the exact installation source for reproducible installs.

DSH compatibility
*
Runtime surfaces
web
Release source
github
Registry updated
10/13/2026

Versions

1.0.0stable
10/13/2026

Related plugins

Loading related plugins…

Latest
1.0.0
DSH
*
HMR
Process restart
Tree shaking
Safe tree shaking not declared
Unpacked size
Unavailable
Files
Unavailable
Surface
web
License
MIT
Source
github
GitHub
★ 0
Weekly downloads
0
Last push
10/13/2026
View source ↗
README badge

Click the badge to copy Markdown for your README.

Do you maintain this Plugin?Claim benefit · Priority security scan

Verify the GitHub repository declared in package.json to manage this listing. After you claim it, Hub will prioritize a security scan of the current version and publish the result when it passes.

Claim this Plugin →
Report an issue

Related plugins

More verified plugins in agents-orchestration.

Hindsight Coding Agents@vectorize-io/hindsight-coding-agentsReflect-only Hindsight long-term memory for coding agents (harness-pluggable: opencode, …), with automatic background ingestion (no setup CLI).Ds Botds-botDS Bot: a team of long-lived Bots for DeepSeek Harness, with Main Bots, Bot-to-Bot messages and group chatsDSCODE Desktop@toddzheng024/dscode-desktopCoding collaboration, memory, browser workflows and task management for Harness Desktop.Headless@deepseek-ai/dsh-headlessThe dsh one-shot bundle: a direct core Agent/Session runner over dsh-base with no Host, HTTP, or browser layer

README

think-level — 会话思考档位,每会话六档推理开关

每个会话页面单独设定推理档位(关闭 / 低 / 中 / 高 / 最高 / 超高),子代理继承所属会话档位。

中文 · English


✨ 功能

  • 每会话独立:会话输入区「思考 ▾」六档圆弧选择器,写入本会话档位,会话之间互不影响
  • 子代理继承:沿 session.header.parentSession 归属链,找到最近单独设定的会话档位
  • 就近收敛:按目标模型 reasoningEfforts 实际支持的档位自动收敛(如四档模型上 中→高、最高→超高),绝不直发模型不认识的档位
  • 模型选择器清理:隐藏 DSH 原生「推理等级 >」入口与其二级面板(能力词典保留给请求链,界面不再露出)
  • 双入口同写:UI 选择器与 set_thinking_level 工具走同一写通道,实时生效、持久化到 profile 补丁层

📦 安装

dsh plugin --profile web add github:373831615/dsh-think-level

本地目录安装:dsh plugin --profile web add <本目录绝对路径>。

🧩 兼容性

项支持
DSH 内核0.2.x —— 实测 0.2.1-alpha.1(开发期同时覆盖 0.2.0-rc.2)
界面DSH Web(client 半 platform: web)
Node.js≥ 20.18.1(安装期)
  • 依赖声明:@deepseek-ai/dsh-settings(可选 peer,内核自带)^0.2.0-rc.1 || ^0.2.1-0、@deepseek-ai/schemastery@^3.18.1
  • 内核 API 基线(均为 0.2.x 世代):agent/request 瀑布、volatile Config、settings 表单写通道、conversation.input.left 插槽、remote.session 命名空间
  • 0.1.x 未测不承诺;0.3.x 未出、届时回归再认证

🧭 使用

六档从低到高(内部档位 id):

思考 ▾关闭低中高最高超高
档位offlowmediumhighxhighmax
  • 会话输入区(模型选择器同排左侧):点「思考 ▾」为本会话选档;未设定的会话显示并跟随默认档
  • 对助手说:「把本会话思考调到超高」→ 工具 set_thinking_level(action=get|set、target=session|default)
  • 默认档(未单独设定的会话生效):set_thinking_level target=default,或补丁层 think-level.config.level

⚙️ 生效机制

「思考 ▾」/ set_thinking_level
        │  (同一写通道:configForms → remote.settings → volatile)
        ▼
Config: level(默认档)+ sessions{会话id: 档位}
        │
        ▼  agent/request 瀑布(prepend,最外层=最终值)
有效档位 = sessions[本会话] ?? 沿父链继承 ?? level
        │  就近收敛到 reasoning.efforts
        ▼
每次模型调用的 reasoningEffort   (子代理继承所属会话档位)
  • 标题生成、压缩摘要等内部调用不走该瀑布(保持便宜)
  • 模型配置里的 reasoningEfforts 声明是档位词典(模型认哪几档的元数据),留给请求链使用;删掉会让六档开关失效

🗂 目录结构

think-level/
├── main.js            # Host:Config(level + sessions)+ agent/request 改写 + set_thinking_level 工具
├── client.js          # Client:会话输入区「思考 ▾」+ 模型选择器推理等级隐藏
├── cordis.patch.yml   # Loader 行(id=think-level,即设置命名空间)
├── locale/            # 显示元数据(en / zh)
├── docs/banner.svg    # 横幅
├── icon.svg           # 图标
└── package.json

🧪 开发备注

  • Client 半(client.js)存盘即生效(client-hmr 轮询 + SSE 热替换,或刷新页面)
  • Host 半(main.js)改码受模块代际缓存约束:同身份热替换在运行中进程里不生效,可靠做法是换包名/目录/行 id 全新身份重装,或重启内核
  • 工具契约坑:output.render(args, value) 首参是工具入参;schemastery z.dict(inner, sKey) 第一参是值 schema
  • 隐藏推理等级入口的刀口:remote 命名空间方法是 getter-only 访问器,包装要打底层 invokeRemote

📄 许可证

MIT