reasoning

dsh-cot-profile

Chloride233

Real-time chain-of-thought trajectory profiling for DeepSeek Harness: live wording indicators, profile-family judgment, per-session measurement records

dsh-custom-provider-settings

supersealwqas

DeepSeek Harness 第三方 API 与自定义模型设置插件：支持请求头、User-Agent、模型列表、图像输入和思考等级 | WebUI plugin for third-party APIs and custom models with request headers, image input, and reasoning levels

dsh-grayprint

HongzhongL

GrayPrint — dsh web plugin for live reasoning-style fingerprints. Scores paragraph openers against grayscale/current references and independently tracks Let me ↔ We need. 94.2% holdout.

dsh-think-autoexpand

2877905731

dsh-think-autoexpand：让 DeepSeek Harness 里的 Think 思考行始终可见，并在流式输出时自动展开，实时查看 Agent 完整推理过程；保留工具卡折叠体验。

dsh-think-window

azazo1

DSH web plugin: cap expanded thinking blocks with a bounded scroll window.

dsh-model-modes

DTSFO

Capability-aware reasoning controls and Fast model routing for DeepSeek Harness

plugin-effort-slider

mrSutivu

Notched reasoning-effort slider grouped with the model picker for DeepSeek Harness — native styling, themeable, i18n

dsh-v4-anchor

Starfie1d1272

Minimal first-turn RL anchor for DeepSeek V4 on DeepSeek Harness Standard.

dsh-model-selector

bitterSmilezzz

DeepSeek Harness (DSH) 的增强模型选择器：单层菜单（搜索 + 分组）+ 底部内联推理强度（Effort）滑杆。

dsh-reasoning

BoWuGit

Codex-style /reasoning command for DeepSeek Harness Web

dsh-matrix-skin

emeryxu1-blip

Black-first Matrix hacker skin for DeepSeek Harness (DSH) Web — live session-powered code rain and readable provider reasoning.

dsh-cot-en2cn

Eyeing0721

DSH Web 插件：展开英文思考块即时呈现中文译文，不改动任何会话原文。

luxueliu-reasoning-efforts

luxueliu

DSH里只有ds能选推理强度？20个常用模型推理强度按钮已就位！涵盖grok/Gemini / Kimi/glm……20个模型仅预设，实际槽位无上限！可以任意添加你的本地网关模型！（非 ds 系网关模型推理强度档位插件 + 路由级 llm-pi-ai 补丁）

dsh-reasoning-cn

William123666

Unofficial DeepSeek Harness plugin for Simplified-Chinese reasoning steering and opt-in translation.

dsh-plugin-effort-declare

zimodzh

为手工 OpenAI 兼容模型补上推理档位声明的 DSH 插件 / DSH plugin to declare reasoning efforts on hand-added OpenAI-compatible models.