HanaAyane/dsh-reasoning-effort
291Last commit Aug 14, 2026
dsh-reasoning-effort DSH plugin
dsh-reasoning-effort adds a slider for reasoning effort selection below the chat input, with levels automatically adapting to the current model. It syncs with the `/model` command and offers a drag-to-adjust, snap-to-value experience. The plugin is default enabled and requires no extra configuration beyond installation.
How to install the dsh-reasoning-effort DSH plugin
dsh plugin --profile web add github:HanaAyane/dsh-reasoning-effort#mainCopying does not run this command. Review the repository and version before installing the dsh-reasoning-effort DSH plugin.
dsh-reasoning-effort DSH plugin data source
dsh-reasoning-effort DSH plugin snapshot date: Aug 16, 2026
discovered
What the dsh-reasoning-effort DSH plugin can do
- Continuous drag slider with snap-to-value on release
- Auto-adapts effort levels from the model's declared reasoning.efforts
- Syncs session model state with DSH's native `/model` command
- Dark/light theme support with gradient indicators
- Graceful fallback: reverts to last confirmed level on update failure
- No additional network telemetry, credential handling, or server-side storage
Where the dsh-reasoning-effort DSH plugin fits
- Quickly adjust reasoning effort for simple Q&A vs. complex tasks
- Switch between predefined effort tiers (e.g., off, high, max) on the fly
- Visually confirm current effort level before sending a message
- Use alongside DSH model selection for fine-grained control
- Enable the 'Big Fat Fish' slider for a playful alternative UI
Who the dsh-reasoning-effort DSH plugin is for
- DeepSeek Harness users who want Codex‑style reasoning effort control
- Developers working with multiple models that support different effort levels
dsh-reasoning-effort DSH plugin limitations
- Requires DeepSeek Harness packages version 0.1.0-rc.6 and Node.js 22.19+
- Only works in the web Profile; other profiles are not supported
- The slider only shows efforts declared by the model; models without declarations will show a message and require manual settings.yaml edits
- Plugin is in early development; upstream UI or service changes may require updates
dsh-reasoning-effort DSH plugin: from the repository README
Quoted from the HanaAyane/dsh-reasoning-effort README, the upstream source of the dsh-reasoning-effort DSH plugin. Copyright remains with the original authors.
**把 Codex 风格的“模型 + 推理强度”控件直接带进 DeepSeek Harness。** [English](README.en.md) · [最新发行版](https://github.com/HanaAyane/dsh-reasoning-effort/releases/latest) · [反馈问题](https://github.com/HanaAyane/dsh-reasoning-effort/issues) [](https://github.com/HanaAyane/dsh-reasoning-effort/tree/main) [](https://github.com/deepseek-ai/deepseek-harness) [](LICENSE) </div> 第一次打开插件时，你会在 DSH 输入框下方看到新的模型入口。点击后，弹层上方是推理强度滑块，档位随当前模型自动适配，下方仍然是熟悉的模型选择入口。插件默认启用，并与 DSH 的 `/model` 命令保持同步。 ## 第一次使用：三步完成 ### 1. 安装插件 #### 让 Agent 安装（推荐） 如果当前 Agent 可以执行终端命令，把下面这段话完整发送给它： ```text 请为 DeepSeek Harness 的 web Profile 安装 dsh-reasoning-effort 插件。 只执行下面两条命令，不要修改其他 Profile： dsh plugin --profile web add github:HanaAyane/dsh-reasoning-effort#main dsh --profile web --dump-config 确认输出中出现 dsh-reasoning-effort 后告诉我安装结果。 不要替我关闭或重启正在运行的 DSH；安装完成后提醒我手动重启 DSH Web Host。 ``` Agent 应当返回安装结果，并明确告诉你配置中是否已经出现 `dsh-reasoning-effort`。 #### 手动安装 也可以自己打开 PowerShell 执行： ```powershell dsh p
Read the full READMERepository license: MIT
dsh-reasoning-effort DSH plugin questions
How do I install dsh-reasoning-effort?
Run `dsh plugin --profile web add github:HanaAyane/dsh-reasoning-effort#main` in PowerShell, then restart DSH Web Host. Use `dsh --profile web --dump-config` to verify the plugin appears in the config. Alternatively, ask your agent to execute the installation commands for you.
I installed the plugin but the slider doesn't show up. What should I do?
First, make sure you restarted DSH Web Host after installation. Then check that the 'Reasoning Effort Slider' toggle is enabled in Settings → General Settings. Also verify the current model declares at least two effort levels in its DSH model directory; if not, see the FAQ about adding manual declarations.
How can I add reasoning effort levels for a model that doesn't have any?
Edit `~/.dsh/settings.yaml` and add a `reasoningEfforts` section under the model entry. For example, for a GLM-5.3 model on zai coding route, specify keys like `low`, `high`, `xhigh` with their API values. After adding, restart DSH Web Host. The plugin will automatically pick up the new levels.
How do I uninstall the plugin?
Run `dsh plugin --profile web remove dsh-reasoning-effort` in PowerShell, then restart DSH Web Host. The native model selector will be restored automatically.
Does the slider send any data to external servers?
No. The plugin does not add any telemetry, credential handling, or server-side storage. All state is local to your browser and DSH instance.