dsh-reasoning-effort-slider
Run the following command in DeepSeek Harness:
dsh plugin install lyaoliu/dsh-reasoning-effort-slider
Paste the following prompt into your AI chat to install this plugin:
Run the plugin install command in DeepSeek Harness to deploy; the source repository is available at https://github.com/lyaoliu/dsh-reasoning-effort-slider
About this plugin
Switching reasoning effort across different models in DSH Web Host usually means repeatedly opening settings.yaml, checking which levels a model supports, and typing in the correct wire values. One small mistake and the server rejects the request. dsh-reasoning-effort-slider collapses all of that into a single slider drag: seven levels from off to max at a glance, with auto-declaration of defaults and hot-reload taking effect immediately, no restart required.
The plugin also ships with self-healing and compatibility fixes. When the server rejects an illegal level, it parses the valid list from the error, rewrites the config, and snaps to the nearest supported level. For pi-ai models it silently sets supportsDeveloperRole to false to avoid role conflicts. A built-in knowledge base handles wire value conversions such as off to none, keeping the front-end slider semantics in sync with the backend protocol. Host logs are emitted when declaration or self-healing fails, making troubleshooting straightforward.
If you frequently tweak reasoning depth across multiple models and are tired of hand-editing config files or chasing incompatibility errors, this plugin is for you. The whale-fall theme and chibi character animations add a touch of charm to what is otherwise a dry tuning task.
Use Cases
- Drag a slider to switch reasoning depth without hand-editing settings.yaml
- Auto-recovers when the server rejects an invalid level by parsing and rewriting the valid list
- Rapidly tweak across multiple models without repeatedly looking up mapping tables
- Hot-reload applies changes instantly, saving restart cycles during prompt tuning
Best For
- DSH Web Host users who frequently switch inference levels across models
- Prompt engineers tired of hand-editing YAML and wire-value mappings
- Developers who prefer a visual UI over plain-text configuration files
Related Plugins
A unified suite combining hot runtime injection, task-aware thinking-mode routing, and a graded session protocol with red-team gates to sustain model diligence across long-horizon inference.
ModLens is a vision plugin for DeepSeek Harness that gives text-only models sight by reading images pasted directly into chat, with zero-config setup and multiple vision engines.
On-demand vision for text-only DeepSeek Harness agents: built-in free keyless vision chain and 14 vision tools, routing image turns as tool calls to vision models with pixel fidelity, no Python needed, one-command install.
Give text-only models in DeepSeek Harness eyes, enabling image Q&A, long-screenshot OCR, UI restoration, and GUI visual tasks.