DeepSeek Harness Plugins · Model Inference
Plugins that extend DeepSeek Harness (DSH) with vision, workflows, memory, web tools and more.
A plugin for DeepSeek Harness providing image understanding, generation, and browser automation.
Adds vision capabilities to text-only models via tool calling, enabling seamless image viewing and interaction.
A zero-dependency skill that turns visual requests into high-quality HTML/SVG artifacts like posters and landing pages.
Adds an Advanced Model Settings section to DeepSeek Harness WebUI, managing LLM provider configurations with unified validation rules.
A model inference tool that provides image description and Q&A capabilities by sending local images to DashScope API.
Displays LLM cost per step and automatically updates price tables.
Automatically identifies DeepSeek API peak and valley pricing times and routes to cheaper or stronger models accordingly.
Adds vision capabilities to DeepSeek Harness text-only main models, automatically invoking configured multimodal models to recognize images and return text descriptions.
Fixes failures from OpenAI-compatible providers (e.g., Snowflake Cortex) that terminate streams without a finish_reason, treating it as a successful finish.
An AI content pipeline for creators with zero external dependencies, featuring topic generation, rewriting, copywriting, and scheduling.
AI fridge manager to recognize ingredient photos and recommend recipes, supporting inventory and shopping alerts.
Leverages local CLI OAuth tokens to serve ChatGPT and Claude subscriptions without separate API keys.