Toukaiteio/dsh-effort-tweak
30Last commit Aug 13, 2026
dsh-effort-tweak DSH plugin
DSH Effort Tweak integrates into the DSH Web UI, adding a 'Reasoning effort' page under Settings. It lists models explicitly configured in `llm-pi-ai.providers.<provider>.models` and allows selecting one of seven effort levels per model. Changes are saved through DSH's settings service, eliminating manual configuration file edits.
How to install the dsh-effort-tweak DSH plugin
dsh plugin --profile web add ./dsh-effort-tweak-0.1.0.tgzCopying does not run this command. Review the repository and version before installing the dsh-effort-tweak DSH plugin.
dsh-effort-tweak DSH plugin data source
dsh-effort-tweak DSH plugin snapshot date: Aug 16, 2026
discovered
What the dsh-effort-tweak DSH plugin can do
- Adds a Settings → Reasoning effort page to the DSH Web UI.
- Lists models explicitly configured in `llm-pi-ai.providers.<provider>.models`.
- Lets you choose from seven effort levels: off, minimal, low, medium, high, xhigh, and max.
- Preserves the model's other fields and existing provider-specific wire values.
- Saves changes through DSH's settings service, avoiding manual configuration-file editing.
Where the dsh-effort-tweak DSH plugin fits
- Tuning reasoning effort for a specific model without editing config files.
- Quickly switching between effort levels for different tasks in the same provider.
- Managing model reasoning effort across multiple models in a custom provider.
- Integrating with DSH's custom provider setup to fine-tune LLM behavior.
Who the dsh-effort-tweak DSH plugin is for
- DSH users who use custom providers with models that support reasoning effort.
- Developers or advanced users who want to control model reasoning without manual config editing.
- Users who need to adjust reasoning effort on the fly through the Web UI.
dsh-effort-tweak DSH plugin limitations
- The page intentionally does not show catalog-only providers; models must be explicitly configured in a custom provider.
- Only one model can be configured at a time.
- Requires a restart of the DSH Web UI after installing or updating the plugin.
- Only works within the `llm-pi-ai` namespace and for models that support reasoning effort levels.
dsh-effort-tweak DSH plugin: from the repository README
Quoted from the Toukaiteio/dsh-effort-tweak README, the upstream source of the dsh-effort-tweak DSH plugin. Copyright remains with the original authors.
English | [中文](README.zh-CN.md) DSH Effort Tweak adds a small settings page for choosing the reasoning levels available to each model in a DeepSeek Harness custom provider. It is designed for the case where the model capability must be edited in `llm-pi-ai` configuration but the official Web UI does not yet expose that control. ## What it does - Adds **Settings → Reasoning effort** to the DSH Web UI. - Lists models explicitly configured in `llm-pi-ai.providers.<provider>.models`. - Lets you choose `off`, `minimal`, `low`, `medium`, `high`, `xhigh`, or `max` for one model at a time. - Preserves the model's other fields and existing provider-specific wire values. - Saves through DSH's settings service, so no manual configuration-file editing is needed. The page intentionally does not show catalog-only providers. Add a model list to a custom provider first, then open this page to tune that model. ## Install from a release ### Windows PowerShell Run PowerShell as your normal user and execute: ```powershell irm https://raw.githubusercontent.com/Toukaiteio/dsh-effort-tweak/main/scripts/Install-DshEffortTweak.ps1 | iex ``` To install into another Profile without starting Web UI:
Read the full READMERepository license: MIT
dsh-effort-tweak DSH plugin questions
How do I install DSH Effort Tweak?
You can install it using the provided PowerShell script on Windows or the bash script on macOS/Linux. The scripts download the latest release, verify the SHA-256 digest, and install it with the `dsh plugin` command. Alternatively, you can build from source with `pnpm install && pnpm build && dsh plugin add ./dsh-effort-tweak-0.1.0.tgz`.
Why don't I see my model in the reasoning effort page?
The page only shows models that are explicitly listed in a custom provider's model list (`llm-pi-ai.providers.<provider>.models`). Catalog-only providers are not shown. Make sure you have added your model to a custom provider's configuration before trying to adjust its reasoning effort.
Can I set multiple models at once?
No, the interface lets you choose the effort level for one model at a time. You must repeat the selection for each model you want to configure. This is by design to keep the interface simple and prevent accidental bulk changes.
What effort levels are available?
The plugin offers seven levels: off, minimal, low, medium, high, xhigh, and max. The `off` level disables reasoning effort, while the others correspond to increasing amounts of reasoning work. The exact behavior depends on the model's support for these levels.
Do I need to restart DSH Web UI after installing?
Yes, after installing or updating the plugin, you must restart the DSH Web UI so the new client bundle is loaded. The install scripts will start the default `web` profile unless you provide the `--no-start` or `-NoStart` flag.