DeepSeek Harness Plugins · Model Inference
Plugins that extend DeepSeek Harness (DSH) with vision, workflows, memory, web tools and more.
A model switch for DSH that points to any OpenAI-compatible endpoint with curated presets and automatic fallback on rate limits.
DSH Effort Tweak adds a settings page for adjusting the reasoning effort levels of models in custom providers for DeepSeek Harness.
dsh-vision-tools enables DeepSeek text models to understand images via vision APIs, supporting paste, drag-and-drop, and file selection.
A minimalist WeChat plugin that supports QR code login and file sending/receiving.
A multi-panel live subagent viewer for DeepSeek Harness with a subagent_with_model tool for per-call model/provider override.
This plugin enables DSH users to stream接入 models via Command Code's Go plan subscription using the /alpha/generate endpoint, with automatic model list synchronization and Reasoning Effort support.
dsh-youmind-plugin is a DeepSeek Harness plugin that integrates YouMind OpenAPI for searching, calling APIs, and querying knowledge bases.
This plugin enables text-only LLMs like DeepSeek to process images in chat by automatically routing them to a vision model for description and feeding the text back, enabling multimodal conversations.
dsh-model-router is a lightweight routing plugin for DeepSeek Harness that adds deterministic keyword routing, a default-off allowlisted model route tool, and an isolated image generation channel.
An adapter plugin that routes DeepSeek Harness's LLM interface to the local Qoder CLI, enabling unconfigured access to built-in and custom models.
An unattended multi-agent research lab for empirical/computational mathematics in economics and finance, integrated into DeepSeek Harness with rigorous checks and verification.
This plugin enables pasting images in DeepSeek Harness, passing them directly if the main model supports vision, or converting them to text via sidecar for pure text models to address the lack of image support.