DeepSeek Harness Plugins · Model Inference
Plugins that extend DeepSeek Harness (DSH) with vision, workflows, memory, web tools and more.
Model Router and Cost Optimizer for DeepSeek Harness that routes simple queries to a cheap model, handles transient failures gracefully, and shows live token usage and cost figures.
dsh-llm-oauth is a standalone OAuth/subscription-plan LLM plugin for DeepSeek Harness, enabling subscription token refresh and multi-provider support.
A custom LLM provider plugin for DeepSeek Harness that integrates Command Code via its own streaming protocol, enabling support for all plans including Go with API key authentication.
Enables ChatGPT login via OAuth to use Codex models within DeepSeek Harness without an OpenAI API key.
Real-time display of DeepSeek account balance in the sidebar, supporting multiple currencies, with settings page for balance trends and usage statistics.
Manages SillyTavern presets, message injection, tool grouping and presets, MCP tool control, and single-file sharing for DeepSeek Harness.
A fun-only plugin: in Empty Fort mode the model performs multiple thinking rounds and tool calls (default 10) but outputs nothing but a single newline.
Tuning Engines CLI is a governed AI runtime for model inference, fine-tuning, and workflow management via a unified API.
A zero-dependency, deterministic statistics tool plugin for descriptive statistics, percentiles, frequency distribution, and correlation calculations.
A one-click draft polisher for the DeepSeek Harness composer, using deepseek-v4-flash to rewrite drafts in the input box, with results replaceable and editable while preserving the original language and structure.
Adds a reasoning effort settings page for custom third-party API providers in DeepSeek Harness, enabling thinking intensity configuration.
This plugin allows using GPT models from a ChatGPT account in DeepSeek Harness without needing an OpenAI API key.