kakadeka/Baton 预览 preview

kakadeka/Baton

Plugin插件 ⭐ 3 Apache-2.0 Sessions & Context会话与上下文

传递你的项目,而非你的上下文。

Project Overview项目介绍

Baton is a project memory and handoff tool built for AI coding agents, supporting Codex, Cursor, Claude Code, and DSH. It stores all project context including tasks, progress, key decisions, known pitfalls, and Git state directly inside your project repository. This lets any AI agent pick up where the last one left off, even when you switch agents, move to a new computer, or start a fresh chat after an old conversation gets too long. Baton offers two installation levels: a one-time user-level install for your computer, and a project-level install that you run once per project.

Baton is designed specifically for users who build long-running projects entirely with AI assistance, rather than writing most code manually. After installation, it adds commands like Baton init and clock in to all supported AI agents, so you can invoke it directly from your AI chat. The recommended workflow is to complete the user-level install once on each computer you use, then initialize Baton in every long-running project to set up the project memory structure. It also automatically checks that all code changes are committed and pushed to the remote Git repository when you clock out of a work session.

Baton is released under the open source Apache-2.0 license, and requires Git and Node.js 18 or newer to work properly. It does not run as a background service, nor is it a cloud storage service, so it never automatically uploads your project data to external servers. It will not modify your existing project code automatically during installation or initialization, and all updates must be triggered explicitly by the user, no background updates occur. If your project contains sensitive information, you should host it on a private Git repository that you control, as Baton does not change repository visibility settings.

Baton 是一款面向 AI 编程工具的开源项目记忆与交接工具,支持 Codex、Cursor、Claude Code 等多款 AI 编码智能体,也可集成至 DSH 使用。它会在本地项目目录内记录所有开发任务进度、重要决策、常见的已知问题和 Git 状态,方便用户切换工具、开启新对话或更换设备时,下一个不同 AI 能从已记录的状态继续开发工作。

它适合不自己写代码、完全依赖 AI 进行长期项目开发的用户。推荐安装方式为先安装用户级 Baton 到本地电脑,再在每个长期项目执行 Baton init 完成项目级初始化。用户切换 AI 工具、跨设备继续开发或者开启新对话时,都可以调用 Baton 读取已有项目记忆,避免重复输入上下文。

Baton 采用 Apache-2.0 开源协议,需要预先安装 Git 和 Node.js 18 以上版本,不依赖后台运行,也不是云服务。它不会自动修改项目代码,也不会上传用户项目到云端,若项目包含敏感信息,推荐用户将项目托管在自己控制的私有 Git 仓库中。更新需要用户主动发起命令,不会在后台自动更新。

Pre-install check安装前体检Compatibility · Security兼容性 · 安全性 2 warnings2 项注意
  • Only 3 stars - very few users, little community feedback星标只有 3,几乎没人在用,遇到问题缺少社区反馈
  • No DSH plugin manifest detected - it may only carry the dsh-plugin topic, so the install method must be confirmed on the spot未检测到 DSH 插件清单:可能只是打了 dsh-plugin 话题,安装方式要现场确认
  • Not DSH-native: a multi-platform tool that may require Node / Electron or another runtime first非 DSH 原生,是多平台兼容工具:可能要先装 Node / Electron 等运行时
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 web add github:kakadeka/Baton

把 kakadeka/Baton 加入你的 DSH 配置(web profile)即可启用。

READMEREADME

Baton — Hand Your Project to the Next AI Without Starting Over

简体中文 | English

Baton is a project memory and handoff tool for Codex, Cursor, and Claude Code. It keeps tasks, progress, important decisions, known pitfalls, and Git state inside your project. When you switch AI tools, start a new chat, or move to another computer, the next AI can continue from the recorded state.

Baton is useful when you do not write code yourself but use AI for a long-running project. You repeat less context, lose fewer unfinished tasks, and can check whether work was really committed and pushed.

Baton is not an AI, a cloud drive, or a cloud service. It is not a background program. It is a set of project rules, templates, and scripts that AI tools read and follow only when you give a Baton command.

When Baton helps

  • Switch AI tools: use Cursor yesterday, then let Codex or Claude Code continue today without explaining everything again.
  • Switch computers: finish at the office, pull the project at home, and bring the progress and warnings with it.
  • Start a new chat: when a conversation becomes too long or loses context, the new chat can read the current task and last handoff.
  • Maintain a long project: keep decisions, rules, and pending work outside temporary chat history.
  • Avoid missed Git steps: after clock out, Baton requires commit and push checks, followed by remote revision verification.
  • Avoid repeated mistakes: save verified pitfalls and long-term decisions so the next AI reads them before changing the project.

Baton cannot guarantee that an AI never makes mistakes. It also does not replace GitHub. It gives project facts a stable home and asks the AI to verify claims with real files and Git results.

Understand the two installation levels

Level How often Where it goes What it does
User level Once on each computer Skill folders for your Codex, Cursor, and Claude Code user Teaches all three AI tools commands such as Baton init and clock in
Project level Once for each project Inside the project folder Creates project memory, handoffs, and entry files that can travel through Git

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 JARVIS-Mission-Control-DeepSeek 下一个 Next dsh-browsercontrol-mcp →