dsh-memory

Identified

Three-layer memory for DeepSeek Harness — a small always-on core block plus on-demand topic files (profile, preferences, decisions, lessons), with source-annotated writes and supersede-not-overwrite h

dsh-memory · JOJO&Adam

DSH 的三层记忆中枢，一个包装齐：

层	解决什么	落点
L1 全局	跨项目、跨会话的画像与偏好	`~/.dsh/memory/`（与 AGENTS.md 协议一致）
L2 项目	单项目防漂移：阶段、决策、约束别丢	项目内 `AGENTS.md` + `.dsh/memory/`
L3 人格	角色的持久人格与抗压缩记忆	可选，接 Stemem MCP

本地优先、零出站。不替代 DSH 原生记忆，只做路由、体检和接线。

安装

方式一：从本地目录（开发）

dsh plugin --profile desktop add link:/absolute/path/to/dsh-memory

方式二：从 npm

dsh plugin --profile web add @jojonadam/dsh-memory        # Web 版
dsh plugin --profile desktop add @jojonadam/dsh-memory    # 桌面版

方式三：只用技能（不装包）

cp -r skills/memory-hub ~/.dsh/skills/

装好后在会话里说「做一次记忆体检」或「帮我建项目记忆」，agent 会按 memory-hub 技能路由。

验证

node test/provider.test.mjs
python3 skills/memory-hub/scripts/mem_health_scan.py --path ~/.dsh/memory/MEMORY.md

包含

lib/index.js                         bundle 入口（注册内置 skill provider）
cordis.patch.yml                     profile patch
skills/memory-hub/SKILL.md           主技能
skills/memory-hub/references/        L1/L2/L3 细则 + Stemem 调用范式
skills/memory-hub/scripts/           健康扫描 / 项目骨架生成（零依赖 Python）

边界

不收集 PII，不外发数据。
人格是计算模拟，不是真实情感。
凭据只进钥匙串，记忆只存指针。

品牌

JOJO & Adam · https://jojo-adam.com

License

MIT（Stemem 运行时 MIT；其底层人格引擎 NeshamaEngine 为 JOJO/Neshama 自有 IP，依原项目许可）。

Comments

Loading…

Similar plugins

dsh-memory-lite

On-demand memory for DeepSeek Harness: session logs are distilled into layered markdown (a summary, an index and per-topic files), and a mem_query tool returns only the matched entries rather than the

Memory & ContextManifest valid

★ 0

dsh plugin --profile web add dsh-memory-lite

dsh-hybrid-memory

by Frog755

Three-layer local memory for DeepSeek Harness: L1 snapshot memory (MEMORY.md/USER.md) read on demand through the memory_* tools + L2 searchable knowledge base (facts + SQLite FTS5, Chinese search via

Tools & CapabilitiesTerminal & ClientsMemory & ContextManifest valid

★ 3

MIT

JavaScript

Sep 10, 2026

dsh plugin --profile web add @frog755/dsh-hybrid-memory

dsh-mnemon

by omdsh-dev

Composable, view-based memory for DeepSeek Harness. Pluggable sources and strategies, with three-tier memory out of the box.

Memory & ContextManifest valid

★ 743

↓ 13.6k/wk

MIT

TypeScript

Oct 28, 2026

dsh plugin --profile web add dsh-mnemon

dsh-memory-layer

by KLRSL

dsh-memory-layer：给 DeepSeek Harness 的「记忆层」——本地单文件 SQLite、触发式召回、五条可解释清理规则、工作流链；零服务、零 API、零遥测。A memory layer for DSH: local SQLite, trigger-driven recall, explainable cleanup, workflow chains.

Sessions & MessagesManifest valid

★ 8

MIT

JavaScript

Oct 25, 2026

dsh plugin --profile web add dsh-memory-layer

dsh-memory-triage

by AtinyFurina

Memory triage for DeepSeek Harness: at the end of each turn it decides which entries to keep and which to archive, routes project-specific ones into a project bucket while global ones stay in the main

Tools & CapabilitiesDevelopment & InfrastructureMemory & ContextManifest valid

★ 0

↓ 117/wk

GPL-3.0

JavaScript

Aug 28, 2026

dsh plugin --profile web add dsh-memory-triage

dsh-memory-layered

by pomelotea-yuzu

Layered (global + per-project) bounded file-backed memory for DeepSeek Harness — model-curated with user intervention, versioned writes, TTL expiry, session search, and an in-chat memory panel

Manifest valid

★ 0

MIT

TypeScript

Aug 24, 2026

dsh plugin --profile web add @deepseek-ai/dsh-memory-layered