Meta-skill for DeepSeek Harness: install, create, localize (Codex→DSH), iterate, and feed fixes back into skills. Includes a curated knowledge base with keyword + local bge-small-zh semantic retrieval, a 7-point write-back checklist, anti-bloat archiving, an optional agent hook plugin, and self-validation tools.
DeepSeek Harness plugin that auto-nudges agents to consult docs or web search when local trial-and-error becomes costly. DeepSeek Harness 插件：当本地试错成本升高时，自动注入检索提示，引导 Agent 查阅文档或联网搜索。
A project-agnostic multi-agent collaboration protocol for the DeepSeek Harness: loop guards that stop runaway agent loops, a six-field handoff format, risk-tiered review routing, and a fixed verify→commit→push→review→close sequence. One skill instead of the same rules copy-pasted into every agent's instructions.
A structured multi-panelist review workflow for the DeepSeek Harness: N panelists score in isolated subagents, blind to each other; a chair reconciles disagreement by anchor match then Δ-level adjudication; an independent critic audits the process. Works for design, code, paper or project review.
Extensible chat-skin monorepo for the DeepSeek Harness Web GUI: a base registry + plug-in skin packages (Feishu-style chat bubbles), plug-and-play without touching chat logic.
DeepSeek Harness 的 Windows 桌面端插件：托盘鲸鱼图标、桌面快捷方式、开机自启直达桌面窗口，一键切换桌面/网页端。 A Windows desktop companion for DeepSeek Harness: system tray (whale) icon, desktop shortcut, login auto-start straight into the desktop window, and one-click desktop/web switching.
Rule-first pre-step skill router for DeepSeek Harness — pours matched skills on high-confidence hits, stays silent otherwise. Deterministic: zero LLM calls, zero token cost until a rule pours, one pour per session, broken YAML never breaks the session. 规则路由 · 高置信命中 · 静默放行 · 零 Token 消耗
