DeepSeek Harness Plugins · Model Inference
Plugins that extend DeepSeek Harness (DSH) with vision, workflows, memory, web tools and more.
Connects Dosh Brain to DeepSeek Harness, enabling natural language search, save, and read notes.
Integrates Musage desktop features into DSH, monitoring 5 AI provider quotas in real-time with auto-switching.
Enables text-only models to process images by forwarding them to a vision model silently.
Provides a settings card for the DSH LLM auto-retry engine, allowing users to adjust retry counts, backoff strategies, and custom retryable codes.
Integrates Command Code Go as a model provider in DeepSeek Harness, supporting both Web and TUI with various models and usage commands.
Lazily loads MCP tool descriptions to reduce token consumption by only showing tools when necessary.
Integrates DIT.ai model catalog into DeepSeek Harness, supporting OpenAI and Anthropic protocols with native model inference.
DeepSeek Harness (DSH) skill-trigger evaluation plugin using an LLM judge to measure routing accuracy, precision, recall, and confusion matrix metrics.
Smart cost guard for DeepSeek API with peak/valley pricing, hard budget caps, off-peak execution, and token optimization.
Forwards web_search calls to selected search backends with sequential fallback.
DeepSeek Harness expense tracker monitoring balance, daily spend, and per-step costs with visual analysis and cost prediction.
Provides offline design recommendations, reviews, and search tools featuring 84 styles and 99 UX guidelines.