DeepSeek Harness Plugins · Model Inference
Plugins that extend DeepSeek Harness (DSH) with vision, workflows, memory, web tools and more.
Registers Agnes AI text-to-image and text-to-video tools in DeepSeek Harness, supporting multi-reference image-to-video with zero dependencies.
A DeepSeek Harness model-inference plugin that provides taste and style review of model outputs.
Manage multiple Codex subscription accounts in a local DSH pool with quota-aware routing and seamless failover when a model's quota is exhausted.
An open-source DeepSeek Harness bundle that turns allowlisted repository context into reviewable Angles launch-video concepts, supporting preview, rendering, and status tracking.
Pure Node.js AI model query plugin providing model listing and detail lookup by ID or name, with no network dependencies.
A free vision recognition plugin for DeepSeek Harness with adaptive long-image tiling, multi-format export, and history management.
A pure Node skin-type inference plugin that returns clear recommendations based on input parameters with no network dependency.
Configure the default provider, model, and reasoning effort for subagents from the Web settings page — no YAML editing needed, with optional live propagation to existing subagents.
A reasoning plugin that returns structured apology methods and suggestions by scenario, implemented in pure Node.js with no network dependencies.
Adapter that routes model inference through a local OpenCode server inside DeepSeek Harness, requiring no API key and supporting streaming text output.
Maps complete OpenAI-compatible streaming responses missing a terminal finish_reason from an error to a normal stop, preventing agent turns from being aborted.
A pure Node.js algorithm timing comparison tool that precisely computes execution time by parameters and returns detailed results, with no network dependency.