Memory and knowledge · DeepSeek Harness plugins — Page 21

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

Native layered memory for DeepSeek Harness with persona files, workspace scope, semantic recall, and background consolidation

将 Floatboat 的提示词工程精华迁移到 DeepSeek Harness：交付哲学、工具纪律、交付真实性契约、安全边界与委派/记忆纪律（作为 agent preset 的 prompt section 注入器）

Hindsight 记忆服务的随行守护：DSH 启动时拉起本地 daemon，DSH 退出时连同嵌入式 PostgreSQL 一起清理干净，并在设置里提供独立页面。支持 Windows 与 Linux。

DSH web client plugin: a sidebar toggle to start/stop the local Hindsight memory daemon (hindsight-embed) for the DeepSeek Harness desktop app.

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.

Recall and re-run the current session's command history with ↑/↓ keys in the DSH Web composer.

QQ 官方机器人接入插件：让用户通过 QQ（私聊 / 群聊 / 频道 @）直接与 harness 完整 agent 对话（工具 / 记忆 / 子代理 / 文件系统 / 安全护栏，与 Web UI 同源）

DeepSeek Harness plugin: multi-tier memory management, semantic retrieval, structured memory, and model-context awareness for infinite context.

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.

DSH 跨会话长期记忆：命名空间隔离 + L1 索引注入（存在性编码）+ 行动验证写入 + 内容级近重复去重 + BM25 全文检索 + 跨命名空间提升 + 自动蒸馏候选 + 溯源/归档/回滚 + 自动维护。

DSH Web GUI 轻量记忆系统：四个大写 Markdown 文件（USER.md / PROJECT.md / WORKLOG.md / CONVENTION.md）+ 两个动作（append 流水 / distill 沉降）。零外部依赖，只用 Node 内置模块；文件可视化、可 diff、可 git 管理。

DSH 简化版记忆插件:SQLite 条目化记忆(标题/重要程度/来源/内容,增删改查)+ 日历与到点提醒(月视图+时间轴+待执行/已完成双tab)+ 定时任务(cron 驱动的进程内调度:到点拉起会话跑提示词,表达式实时预览/立即运行/执行记录/启停),设置页记忆/日历/定时三子页,中英双语自动跟随。内置:记忆标题经 systemPrompt.context 动态注入为 durable user-role 快照(全局+当前项目,CRUD 后自动刷新),比静态 section 更省 token。

Living memory for DeepSeek Harness — self-tending knowledge base: nightly patrol, temporal decay, RRF hybrid recall, knowledge graph, telemetry. Built by 暖暖 (NuanNuan).

Popup-on-open + periodic reminder to save DSH session logs, with in-popup interval (10min-3h), backup folder & format (fish=raw .zstd / human=readable .txt) settings and one-click incremental copy. / 打开 Web 即弹窗 + 定时提醒保存 DSH 会话日志，弹窗内可设提醒间隔（10分钟-3小时）、备份文件夹与格式（鱼话版=原始压缩 / 人话版=可读聊天记录），一键增量复制。

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).

Cross-session persistent memory for the DeepSeek Harness: a model-facing memory tool, an always-on index section, and a bundled maintenance skill.

Codex-style per-session memory for DeepSeek Harness: qualified sessions auto-distill into per-session drafts (rollout_summaries); manual add/change/forget runs only on explicit user request; layered disclosure (summary + registry); idempotent integration watermark. / 一会话一草稿记忆插件：合格会话自动提炼草稿；手动增/改/忘仅在显式请求时执行；分层披露（总纲+注册表）；幂等整合水印。

Retrieved ≠ injected: CBDC-gated memory for DeepSeek Harness — decides how memory is USED (use/verify/ignore decisions, feedback learning, full audit); local SQLite + FTS5, bounded, no extra model call

Memory protocol enforcement plugin for DeepSeek Harness — forces memory_weave before tool calls, auto-ingests session turns, injects memory context into agent steps. Bridges the opencode-memory MCP server with a hard protocol gate.

给 DeepSeek Harness 补上时间轴记忆：定宽只追加的事实日志 + agent 自己维护的摘要金字塔，固定阅读预算下 recall 成本不随记忆量增长。

Local, cross-session full-text search for DeepSeek Harness conversations: SQLite FTS5 indexing with Chinese word segmentation and accurate message-level jump targeting

dsh自动进化记忆系统