DeepSeek Harness 플러그인 — 371 페이지

自包含的 DSH 简历工作流插件：注册四个 Skill，自带解析器、编辑器与模板，独立安装即可用（事实库仍留在使用者自己的目录）。

简历专家插件 × DeepSeek Harness 适配层（Phase 1：工具化接入）

Headless one-shot session resume for DeepSeek Harness: dsh --profile headless --resume <session-id> "task"

让 DeepSeek Harness 自动重试商汤 429/QUOTA 等瞬时网关故障直到任务完成——启动时给所有 llm-pi-ai provider 热注入加固版 retryPolicy，再也不用手动发「继续」。

Show DSH's hidden injected-context rows (AGENTS.md notices, skill catalogs, recalled sessions, runtime snapshots) back in the Chat transcript, with a per-kind opt-out.

DeepSeek Harness 现代化对话回退与重试插件：纯 UI 图形交互、原地 Prompt 微调、支持工作区与外部文件双引擎安全恢复。

Engineering-discipline skill pack for DeepSeek Harness — code review, simplification, plan-then-execute, test-first, and conflict resolution, delivered as a bundled skill provider plugin.

DSH 一键复活：重启后给所有被打断的会话自动发送「继续」指令（/revive 命令 + revive_sessions 工具 + 浏览器一键按钮）

Undo what an agent did to a directory: content-addressed snapshots, one-command rewind, and an undo you can undo. Works in any folder — git or not.

DeepSeek Harness Artifact/Deliverable plugin: publish workspace files as chat artifacts with inline image previews and download cards.

Rich branching survey system for DeepSeek Harness (DSH) Web GUI: the ask_survey tool extends the flat ask_user_question flow — up to 150 questions, per-option branching (select C -> a different range of questions follows), delayed hover insights with sources, expandable Mermaid diagrams, multi-select, free-text, quick mode (one-click whole-survey answer templates), and reroll/push/discuss pre-flight actions. Host-authoritative pending surveys, rendered as a wizard in the same conversation composer seat.

RigorQuant for DeepSeek Harness: session-scoped unattended, context-isolated multi-agent research for empirical/computational mathematics with a four-part pre-implementation check battery and a jacobian/Lean escalation lane.

Provenance-aware execution security for DeepSeek Harness. Track sensitive data across tool calls and stop risky side effects before execution.

Heads-up Texas Hold'em training for DeepSeek Harness, with an AI opponent, persistent public-action memory, and decision reviews.

Role presets and a model pool for dsh: role-routed delegation with capability conditions and priority fallback

DSH agent preset 分发包：V4 Flash 神模式（opencode-go）——启动时自动把 router-flash preset 安装到 ~/.dsh/.agent-presets。

Per-request tier routing on the Laya judge, plus an Auto tier chip in the composer

三段式 RP 输入面板：台词/动作/心理分框填写，按模板组装后发送

Local memory and skill evolution for DeepSeek Harness

RSS/Atom feed subscription, scheduled fetching, deduplication, LLM-powered summarization, and daily Markdown digests — as a dsh (DeepSeek Harness) bundle.

DSH 规则执行引擎 v3：容器解析 AGENTS.md + 理解器 + 匹配机 + 执行框架

规则透镜（RuleScope Lens）—— DeepSeek Harness 本地插件：AGENTS.md / rules 分层规则加载可视化、64KB 预算治理、写回防护与 FORBID 硬拦截、规则 Lint 与遵守率统计 | Rule visibility, budget governance & hard guards for DSH

DSH rules plugin: activate rule prompts / markdown documents by glob-matching the files an agent reads or edits (Claude Code rules.md style).

DSH × RunningHub 工作流适配插件：让模型直接生图 / 生视频 / 生音频。两个模型侧工具（runninghub_search 发现 + runninghub_call 执行）覆盖「AI 辅助配置工作流 → 读提示词优化文档 → 后台跑工作流 → 结果图/视频回到聊天」全链路；多 Key 池按国内 / 海外分池、额度不足自动换号；数据与工作流配置落在 DSH_HOME/runninghub，自带浏览器配置面板（工作流默认折叠、点开才展开节点）。