better-model-provider
Run the following command in DeepSeek Harness:
dsh plugin install sanshanya/better-model-provider
Paste the following prompt into your AI chat to install this plugin:
To install the better-model-provider plugin in DeepSeek Harness, run `dsh plugin install sanshanya/better-model-provider`; the full source code is available at https://github.com/sanshanya/better-model-provider.
About this plugin
In DeepSeek Harness, key model capabilities like reasoning effort levels and input modalities are hidden in YAML configurations, leading to issues such as no effort control in the picker or image sessions being refused due to missing declarations. The better-model-provider plugin addresses this by exposing these settings in the user interface, making it easy to configure model behavior without delving into code.
Its core functionality is per-model capability editing. Users can customize reasoning effort levels (including toggling specific levels), set input modalities (e.g., enabling image support), and adjust context window and max token capacities. For custom models, it edits declarations directly; for official models, it applies sparse overrides while keeping provider configurations in sync with official updates, ensuring a smooth experience.
This plugin is ideal for developers fine-tuning model responses or end-users needing specific features like image analysis. Compatible with dsh 0.1.x, it allows quick adjustments via an intuitive interface, enhancing DeepSeek Harness's customizability and user experience without complex setup.
Screenshots
Use Cases
- Adjust model reasoning effort levels to optimize response depth
- Enable image input support to prevent model refusal of pictures
- Manage capability override settings for official models
Best For
- Regular users of DeepSeek Harness
- Developers needing to customize model behavior
- Researchers using image analysis features
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.