Agent 협업 · DeepSeek Harness 플러그인 — 125 페이지

Offloads large, low-judgement DeepSeek Harness tool results to small one-shot worker agents before they enter the main model context

Reversible DOM diagnostics and scoped UI repairs for DeepSeek Harness plugins

Mine durable project-rule candidates from user corrections in DeepSeek Harness sessions

Authenticated, policy-gated WebFetchProvider for the DeepSeek Harness web capability seam (ctx.web)

dsh「专家团」bundle：一句自然语言自动组建/持久化一支 12 角色多智能体团队，共享工作区协议 + 阶段门控编排 + 结构化交接 + 质量门禁/自动调度，实现者直接改代码并产出持久工件；带 live 团队浮层（质量门禁/覆盖率/工件预览）。 · Role-based multi-agent expert team for DeepSeek Harness: one sentence in, a staged and gated team delivery out.

DSH Web plugin: completion chime and tab-title alert when agent turns or background jobs finish

定锚（Anchor）· DeepSeek Harness 会话指令定锚插件：为新会话注入一段可自由组合的指令，并在会话被压缩或过长后自动重锚同一段原文，避免人设与规则被挤出上下文。

Remote workspace for DeepSeek Harness: SSH host vault (groups, jump chains, proxy, encrypted credentials), interactive SSH terminal, SFTP explorer, remote workspace binding, and session-scoped remote agent tools.

Harness-Memory-Evolution (HME): cross-session long-term memory for DeepSeek Harness — bounded core memory plus a tag-indexed, self-consolidating archive.

DeepSeek Harness plugin: installs the 'wsl' agent preset — the Windows WSL variant of the standard preset that routes the bash tool through Windows Subsystem for Linux (WSL), with automatic shell discovery and sandbox-aware gating.

AI coding delivery trust layer for governance-aware coding agents — zcode plugin packaging. The plugin surface lives in skills/ and commands/; this file only provides the npm package identity expected by the zcode plugin host.

Turn-level safety inspection, change attribution, safe rewind, and isolated retry for DeepSeek Harness coding sessions.

Agent loop rate limiter — prevents TPM/RPM limit violations by intercepting the LLM streaming pipeline and adding adaptive delays between requests

Per-agent per-step ToolRuntime call limits for DeepSeek Harness

DSH plugin developer sandbox: spawn isolated DeepSeek Harness web mirror instances (separate DSH_HOME, separate port, separate profile) that mount the plugin under development — with optional host API/model inheritance so the mirror can chat directly. GUI panel + agent tools (sandbox_list/start/stop/destroy/logs/build).

DeepSeek Harness plugin: dispatch work to DSH agents from Claude Code / Codex, as native subagents with live progress

Noema long-term memory plugin for DSH: durable, inspectable agent memory with recall tools and a settings page.

Rick and Morty for the dsh web GUI: three-look skin (C-137 portal green / lab coat / Meeseeks blue) plus a desktop-pet companion with six characters, bilingual lines, agent-driven moods, combo petting, and a progression panel

每日工作日志：聚合 Git 提交与本地 agent 对话（deepseek/claude/codex），按自定义模板生成日报/周报/月报；agent 工具按需注入，默认只挂一个 daily_log 派发器。

dsh 工作台插件：聚合日常功能入口的骨架包，待填充 UI。图标统一使用 lucide-react。

dsh 记忆与进化插件：跨会话记住你的偏好与项目习惯（全局 / 项目双作用域），自动注入、自动整理，设置面板一级「记忆」分区可视化管理。

dsh 便签插件：独立便签板 UI（侧栏入口 + 中间列面板，Markdown 编辑器，Typert 直连 host 持久化）

Out-of-tree DeepSeek Harness bundle: an executable implementation of the Harness Engineering Methodology (HARNESS-METHODOLOGY-SPEC v1.0) — agent-discipline prompt section + artifact scaffolding tool + compliance audit tool.

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.