DeepSeek Harness Plugin Hub

Publish and manage complete Harness Profiles. Discover Plugins for your next setup.

Explore

PluginsPresetsDocsNews

Community

Publish a pluginContactReport an issue

Resources

Plugin Hub on GitHubDeepSeek HarnessSystem statusPrivacy notice
© 2026 DeepSeek Harness Plugin HubPowered byPaxTech

Independent and unofficial. Not affiliated with, authorized by, or endorsed by DeepSeek.

Okf Memory — DSH Plugin for DeepSeek Harness
DeepSeek Harness Plugin Hub
ProfilesPluginsCategoriesNewsDocsSign inManage Profiles
ProfilesPluginsCategoriesNewsDocsSign in
← Plugins
O

dsh-okf-memory

Okf Memory

Session Memory → OKF Knowledge Consolidation Plugin: A predictive, neuro-self-learning memory system. Consolidates high-value content from sessions into long-term memory according to the OKF v0.1 specification and automatically recalls it across sessions. Session-to-OKF memory plugin with neuro-self

The plugin will be installed here. Keep web if you are unsure.

npx -y @deepseek-ai/dsh plugin --profile web add github:ZHI-QI/dsh-okf-memory#943c5ddb82572a9b89ee9d30240a65c705d6a510
READMECompatibilityVersions

Description

Session Memory → OKF Knowledge Consolidation Plugin: A predictive, neuro-self-learning memory system. Consolidates high-value content from sessions into long-term memory according to the OKF v0.1 specification and automatically recalls it across sessions. Session-to-OKF memory plugin with neuro-self-learning (predictive recall, uncertainty-driven capture, reinforcement feedback).

Compatibility and provenance

Okf Memory is published as dsh-okf-memory and currently resolves to version 0.1.0. The Hub verifies its manifest and preserves the exact installation source for reproducible installs.

DSH compatibility
*
Runtime surfaces
any
Release source
github
Registry updated
8/27/2026

Versions

0.1.0stable
8/27/2026

Related plugins

Loading related plugins…

Latest
0.1.0
DSH
*
HMR
Process restart
Tree shaking
Safe tree shaking not declared
Unpacked size
Unavailable
Files
Unavailable
Surface
any
License
MIT
Source
github
GitHub
★ 2
Weekly downloads
0
Last push
9/1/2026
View source ↗
README badge

Click the badge to copy Markdown for your README.

Do you maintain this Plugin?Claim benefit · Priority security scan

Verify the GitHub repository declared in package.json to manage this listing. After you claim it, Hub will prioritize a security scan of the current version and publish the result when it passes.

Claim this Plugin →
Report an issue

Related plugins

More verified plugins in memory-context.

Memsearch Dsh@zilliz/memsearch-dshMemSearch plugin for DeepSeek Harness: shared markdown memory across agents, with capture, pre-step context injection, memory-recall skill, and a skill-candidate review panel.Reme@agentscope-ai/remeReMe client and memory integrations for TypeScript agentsStratagate Dshstratagate-dshRecent conversations stay vivid. Older ones fade into summaries, not oblivion. StrataGate gives DeepSeek Harness six-layer, time-decaying memory, while lasting events and relationships settle into a knowledge graph. Bring your memories from other AIs withMeow Memorymeow-memoryCross-session project memory for DeepSeek Harness: seven-layer SQLite memory, first-turn snapshot injection, per-message keyword hits, memory_remember/search/project tools, automatic reflection with reflection-fold UI, and idle-triggered dream consolidati

README

dsh-okf-memory

English | 简体中文

会话记忆 → OKF 知识沉淀插件(神经自我学习驱动)

把会话中高价值内容按 OKF v0.1 规范自动沉淀为长期记忆,跨会话自动唤起。Agent 越用越准:每次选择、跳过、纠错都是学习信号,记忆权重持续更新。

Session-to-OKF memory plugin with neuro-self-learning: predictive recall, uncertainty-driven capture, reinforcement feedback, consolidation & forgetting.

dshfind

特性

  • 记忆四阶段闭环:捕获 → 概念化 → 沉淀 → 唤起
  • OKF v0.1 合规:每个概念是标准 Markdown 文档(frontmatter 硬要求 type),index.md 渐进式目录 + log.md 变更历史,交叉链接用包内绝对路径
  • 神经自我学习驱动:预测性唤起(先预测再检索校验)、不确定性量化(置信低扩大探索)、预测误差驱动捕获(用户纠正/首次披露触发写入)、权重衰减+归档(巩固与遗忘)
  • 强化反馈回路:score = relevance × weight × recency,用户选中候选权重↑、跳过权重↓
  • 技术选型记忆(TechChoice):前端/后端/语言/方案/配置 按维度沉淀候选表 + 当前使用;三档选择规则(多候选展示、单候选直用、领域命中按维度)
  • 写入许可门:type 合法性 → 去重(互补不复制,互建交叉链接)→ OKF 符合性校验

安装

# 任意 profile(如 web):
dsh plugin --profile web add dsh-okf-memory
# 或从本地路径:
dsh plugin --profile web add ./dsh-okf-memory
# 或从 GitHub(拉取源码,需 prepare 构建并向用户授权构建):
dsh plugin --profile web add github:ZHI-QI/dsh-okf-memory
# 或发布到 npm 后免构建授权安装:
dsh plugin --profile web add dsh-okf-memory

插件零运行时依赖(peer 依赖 @deepseek-ai/cordis 由 dsh 运行时提供),安装即用,无需构建脚本。

如何使用

装好后你无需手敲命令。插件会给 Agent 注入一段「记忆纪律」系统提示,让它在会话里自主判断该记什么、该查什么,并调用下面的工具完成。你也可以随时显式地说「记住XX」或「查一下记忆里关于XX的」来主动触发。

4 个工具一览

工具作用什么时候用
okf_remember写入一条记忆(自动去重、校验、落盘)有值得沉淀的新知识时
okf_search按关键词召回,按权重/近因排序开场预取、回答前找相关记忆
okf_read精读某条全量(含交叉链接),并记录一次使用反馈需要完整细节时
okf_forget撤回一条记忆记错 / 不需要时

让它记住(写入)

  • 自动(推荐):你在对话里披露新事实、拍板决策、纠正 Agent、提到技术选型时,Agent 会自己判断是否值得记,不用你开口。
  • 手动:直接说「记住…」即可触发,例如 记住,我的三家门店是韶山/湘乡/塘厦,共用局域网共享文件夹。

什么才算「值得记」:新背景事实/偏好、决策及理由、可复用方法论/流程/经验教训、用户纠正、被确认的反直觉结论、技术选型。 不记:寒暄、单轮临时任务、已有记忆的重复内容、未验证的猜测(猜测归入 Idea,等成熟再沉淀)。

让它回忆(召回)

  • 你问相关问题时,Agent 会先 okf_search 召回再作答。
  • 也可显式说「查一下记忆里关于XX的」。
  • 检索不到会明确告诉你「记忆库没有」,不会编造。

技术选型怎么用(TechChoice)

针对前端/后端/语言/方案/配置这类选型,插件按「三档规则」处理(细节见下文「技术选型三档规则」专节):命中 2+ 候选 → 全部展示给你选;命中 1 个 → 直接用;你未指定技术但命中维度关键词(如「前端」)→ 按该维度记忆处理;你提出新方案/切换/配置 → 追加式更新,不覆盖旧候选。

示例:怎么记、怎么查

// ① 记一条门店事实(Fact)
用户: 记住,我的三家门店是韶山/湘乡/塘厦,共用局域网共享文件夹
Agent: okf_remember(title="门店布局", type="Fact",
        content="# 核心\n\n三家门店共用局域网共享文件夹…", tags=["门店"])
       → 已沉淀记忆 fact/门店布局

// ② 记住前端方案(TechChoice)
用户: 前端就用 React 18 + Vite 吧
Agent: okf_remember(type="TechChoice", title="前端方案",
        content="## Options\n\n| 候选 | 状态 |\n|---|---|\n| React 18 + Vite | active |",
        tags=["前端","技术选型"])

// ③ 问数据库时先召回(而不是去翻本地文件)
用户: 帮我查询数据库
Agent: okf_search(query="查询数据库")
       → 命中「鼎赞数据统一用 mcp-dezensaas-mysql」
       → 按该记忆走 mcp-dezensaas-mysql 服务

记忆库结构

默认 ~/.dsh/memory/(环境变量 OKF_MEMORY_ROOT 覆盖):

~/.dsh/memory/
├── index.md              ← 渐进式目录(okf_version: "0.1")
├── log.md                ← 变更历史(## YYYY-MM-DD)
├── fact/                 ← Fact 背景事实
├── preference/           ← Preference 用户偏好
├── decision/             ← Decision 决策(三段式:数据/分析/结论)
├── method/               ← Method 方法论
├── insight/              ← Insight 洞察
├── idea/                 ← Idea 未成型灵感
├── lesson/               ← Lesson 经验教训
├── techchoice/           ← TechChoice 技术选型(Options 候选表 + Active)
└── .meta/weights.json    ← 学习权重元数据(不污染 OKF 符合性)

技术选型三档规则(用户既定协议)

  1. 命中 2+ 候选 → 全部展示给用户选择,不擅自决定
  2. 命中 1 个候选 → 直接使用
  3. 用户未指定技术但命中维度关键词(如"前端")→ 按该维度记忆处理
  4. 用户说出新技术/切换/配置 → 追加式更新,不覆盖旧候选(保留 v1→vN 迭代轨迹)

配置

项方式默认
记忆库根目录环境变量 OKF_MEMORY_ROOT 或 settings okfMemory.root~/.dsh/memory/
学习参数lib/learning.js 中 PARAMS(衰减天数/归档阈值等)见文件

开发与测试

node scripts/smoke.js        # 核心模块功能验证(19 项)
node scripts/integration.js  # mock dsh ctx 集成验证(24 项)

License

MIT