sens-io/memobranch 预览 preview

sens-io/memobranch

Git-native, auditable long-term memory for AI agents

Project Overview项目介绍

MemoBranch is a local-first long-term memory layer designed for AI agents and production workloads. It organizes conversational evidence, candidate knowledge, and canonical memories into a human-readable Markdown Wiki while using Git as the underlying versioned store, giving every write attribution, rollback history, and cross-machine synchronization. The repository ships an official DeepSeek Harness Cordis plugin that registers memory, wiki, and lint namespaces inside a DSH client, and it also exposes a standard MCP endpoint so Claude Code, Cursor, and any other MCP-speaking agent can attach without modification.

A typical loop is Ingest, then Query, then Lint: raw tool results are filed as immutable Evidence, approved candidates become Wiki Memory pages, and a semantic Linter later checks citations and consistency. Ordinary read queries stay side-effect free; answering filing and Lint repairs need explicit authorization, while model-driven compilation, QA, and semantic enrichment require an LLM API that the operator selects. The project targets engineering teams and individual agent builders who need auditable, portable, Git-versioned memory instead of an opaque vector store.

Runtime requirements are Node.js 20 or newer with TypeScript 6.x, plus a local Git repository acting as the source of truth; remote LLMs are strictly optional for non-model workflows. The codebase is MIT-licensed, specifications and verification reports live under openspec/, and first-run setup means initializing the Git backing store, configuring tenant boundaries, generating encryption keys, and reviewing the documented threat model before any production exposure, since local acceptance evidence does not automatically certify hosted CI, real-model quality, or live deployment safety.

MemoBranch 是面向 AI 代理的本地优先长期记忆层,将对话证据、候选知识与权威记忆组织为可读的 Markdown Wiki,并依托 Git 实现版本追溯、回滚与跨机器同步。仓库自带 DeepSeek Harness 原生 Cordis 插件,按官方文档安装到 DSH 客户端即可注册 memory、wiki、lint 等命名空间工具,同时暴露标准 MCP 接口,Claude Code、Cursor 等遵循 MCP 协议的代理也能直接接入。

典型工作流是 Ingest → Query → Lint:把工具结果沉淀为只读证据,再批准为 Wiki 页面,最后做语义体检与引用校验。普通查询只读,回答归档与 Lint 修复需要显式授权,模型驱动的编译、问答、语义分析才需要配置聊天模型 API。它适合需要可审计、可版本化、可迁移记忆的研发团队与个人代理开发者。

依赖 Node.js 20+ 与 TypeScript 6.x,使用本地 Git 仓库做持久化,可选用远程 LLM 增强;不依赖模型仍可完成归档、检索与回滚。项目以 MIT 协议开源,所有规格与验收记录归档于 openspec/ 目录;首次运行需初始化 Git 存储、配置租户与密钥,并参考威胁模型补齐磁盘加密与最小权限,生产部署前应查阅本地验证记录而非默认信任 CI 结果。

Pre-install check安装前体检Compatibility · Security兼容性 · 安全性 1 warning1 项注意
  • Only 2 stars - very few users, little community feedback星标只有 2,几乎没人在用,遇到问题缺少社区反馈
DSH walks through these 9 checksDSH 会逐条核对这 9 项

Compatibility兼容性

  • DSH, Node, OS and profile requirementsDSH 版本 / Node 版本 / 操作系统 / profile 是否满足要求
  • External dependencies and runtimes (Electron / Python / Docker, ...)外部依赖与运行时(Electron / Python / Docker 等)是否齐备
  • Conflicts with installed plugins: command names, skill / tool names, ports, duplicate MCP registration与已装插件是否冲突:命令名、skill / tool 重名、端口占用、重复 MCP 注册

Security安全性

  • Repo matches the facts registered here; archived or abandoned?仓库是否与页面登记一致,是否归档或长期停更
  • Safety of preinstall / install / postinstall and install.sh / setup.ps1preinstall / install / postinstall 与 install.sh、setup.ps1 是否安全
  • curl|bash, download-then-execute, obfuscation, unrelated domains → stop immediatelycurl|bash、下载即执行、混淆代码、无关域名 → 立刻停止
  • Typosquatting or unmaintained packages among the new dependencies新增依赖里有没有 typosquatting 或无人维护的包
  • Requested permissions vs. what the feature actually needs申请了哪些权限、是否超出功能所需(filesystem / network / shell / clipboard)
  • Any sudo / admin requirement, plus uninstall and rollback是否要求 sudo / 管理员权限,以及卸载与回滚方式

Anything uncertain must be marked unknown with a note on how to confirm it. This site's signal screen is a static snapshot, not a security audit.拿不准的必须标「未知」并说明要我怎么确认。本站的信号筛查是静态快照,不能替代安全审计。

Or use CLI install (for developers)或使用命令行安装(适合开发者)

CLI Install命令行安装

dsh plugin --profile personal-agent add memobranch

把 sens-io/memobranch 加入你的 DSH 配置(web profile)即可启用。

READMEREADME

MemoBranch Logo

MemoBranch

Memory that branches with your agents.

English · 简体中文

Auditable, searchable, portable long-term memory for AI agents

Markdown is the source of truth · Git tracks every change · LLMs provide optional enhancements

Version 1.1.0 Node.js 20+ TypeScript Git native MCP ready DeepSeek Harness plugin CI Local verification documented MIT License GitHub Stars

Why MemoBranch • Capabilities • Quick Start • Web Console • Architecture • DeepSeek Harness • MCP • Operations


MemoBranch is a local-first long-term memory layer designed for AI agents and production use cases. It organizes conversational evidence, candidate knowledge, and canonical memories into a human-readable Markdown Wiki, with Git providing versioning, attribution, rollback, and cross-machine synchronization.

Showing the opening section of the README — the full document lives in the repository以上为 README 开头摘要,完整文档在仓库内 · View the full README on GitHub →在 GitHub 查看完整 README →

← 上一个 Prev dshctl 下一个 Next dsh-ui-beautify →