dsh-reasoning-settings
Run the following command in DeepSeek Harness:
dsh plugin install JuneLearn/dsh-reasoning-settings
Paste the following prompt into your AI chat to install this plugin:
In DeepSeek Harness, you can install this plugin by running the command dsh plugin install JuneLearn/dsh-reasoning-settings, with the complete source code available at https://github.com/JuneLearn/dsh-reasoning-settings.
About this plugin
In DeepSeek Harness, the reasoning effort (thinking intensity) settings are officially limited to first-party DeepSeek models, leaving many users who rely on third-party API providers without this customization. This plugin bridges that gap by adding an independent settings page for custom llm-pi-ai providers, enabling flexible configuration of thinking intensity levels for each model.
Its core capabilities include auto-reading added providers and models, allowing declaration of effort levels from off to max, and customizing the actual values sent to the API for each level. The plugin also fixes subagent inheritance issues, ensuring that reasoning effort is correctly propagated when delegating tasks to child agents.
Designed for developers, researchers, or advanced users of DeepSeek Harness who work with third-party APIs, this plugin is ideal for anyone managing custom model providers. Whether you use relay services for OpenAI-compatible APIs or handle multiple custom models, it provides fine-grained control over reasoning processes to enhance workflow consistency and efficiency.
Screenshots
Use Cases
- Configure independent thinking intensity levels for multiple custom model providers
- Inherit and adjust parent agent's reasoning effort settings in subagent tasks
- Optimize reasoning request formats for third-party relay APIs
Best For
- Developers using DeepSeek Harness with integrated third-party APIs
- Researchers needing to manage multi-model reasoning configurations
- Advanced users or teams seeking fine-grained reasoning control
Related Plugins
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.
An armor-breaking plugin for DeepSeek: rewrites instruction priority, expands refusal-blocked phrases, persists across sessions, and shows a green active indicator.