記憶與知識庫 · DeepSeek Harness 插件 — 第 21 頁

Layered memory, mood, and bounded personality state for DSH.

提供分层记忆管理与背景整合功能

将 Floatboat 的提示词工程精华迁移到 DeepSeek Harness，作为 Agent 预设的 Prompt Section 注入器。

提供本地记忆服务守护，自动管理进程与数据库生命周期。

DSH Web 客户端插件：用于启动/停止 DeepSeek Harness 桌面端本地 Hindsight 记忆守护进程（hindsight-embed）的侧边栏开关

给「大肥鱼」这类养成型 QQ 伴侣用的记忆层：事件编码、情绪加权遗忘、联想召回、关系与技能，还会承认自己记不清。Memory layers for a long-running companion bot.

DSH Web 键盘上下键历史命令回溯插件

QQ 官方机器人接入插件：支持用户通过 QQ 私聊、群聊或频道直接与完整的 DSH Agent 进行同源对话。

提供多级内存管理与语义检索的无限上下文

DeepSeek Harness plugin: one chat/customer-support bot across QQ (SnowLuma), official QQ Bot, and Discord, with per-conversation domains, shared knowledge, and unified push.

Modular QQ character integration kit for DSH: NapCatQQ/SnowLuma, Bilibili, GPT-SoVITS voice, revisable memory and autonomy. Companion services required.

提供跨会话长期记忆管理与全文检索系统。

DSH Web 轻量记忆系统：基于四个 Markdown 文件进行流水记录与沉淀，零外部依赖且可视化。

SQLite条目化记忆、日历提醒与定时任务

提供自维护的知识库，支持混合检索与知识图谱。

DSH 会话日志定时提醒与增量备份插件

Keeps a long agent turn answerable to the user's actual request: a standing maintainer-document reminder, two pre-execute gates (read a same-basename precedent before editing infrastructure; say what a side-effecting step is for), an objective anchor that quotes the user's latest instruction plus an explicit precedence rule, and a nudge stage that corrects before it blocks. Details: (1) A standing system-prompt section that tells the model to read the workspace maintainer documents (plan.md / conventions.md / stack.md / state.md / maintainer/README.md) before every operation, so small-model context compaction cannot erase long-term project memory. (2) A `tools/pre-execute` precedent gate that denies a write/edit inside a guarded infrastructure area until a working same-basename precedent has been read. (3) A `tools/pre-execute` intent gate that polices the turn's side-effecting calls when the model never said what the step is for. (4) An OBJECTIVE ANCHOR: the user's latest instruction, quoted verbatim in a second prompt section with an explicit precedence rule (user's latest instruction > your own last stated plan > a lead you found yourself), so a long turn cannot silently redefine its own task. (5) A NUDGE stage that corrects before it blocks — the first unexplained action only injects a reminder the model reads at its next step, a drift reminder fires once per long turn, and only a repeated offence is denied. Plus a right-sidebar tab that views and edits those documents, and ten runtime knobs exposed in the harness settings page (设置 → 插件 → 插件配置).

DSH 跨会话持久记忆插件，包含模型端记忆工具、常驻索引区和捆绑的维护技能，支持项目级或全局隔离。

提供一会话一草稿的Codex风格会话记忆管理。

提供基于本地 SQLite 的记忆检索与决策网关

强制执行记忆协议并自动注入与存档记忆上下文

时间轴事实日志+摘要金字塔记忆，在固定预算下持续召回历史

用于 DeepSeek Harness 会话的本地、跨会话全文搜索插件：基于 SQLite FTS5 索引，支持中文分词与精确的消息级跳转定位。

DSH 自动进化记忆系统：提供记忆标签页以查看和沉淀记忆，支持“--”快捷指令引导 AI 按需调用 memory_read 工具读取分组记忆。