Picture published by the creator on GitHub.
Keep the reasoning behind every commit an agent makes
Made by Adib Mohsin
View profile on GitHub (opens in a new tab)A desktop app that treats agent sessions as something worth version-controlling.
Every agent run produces checkpoints, and each checkpoint links a commit back to the session that made it — the prompts, the tool calls and the reasoning — so months later you can ask why a change was made and get an answer. It runs Claude Code, Codex, its own agent and anything from the ACP registry side by side against one codebase, with a shared memory so switching agents mid-task does not mean re-explaining everything. Notes in the project, plus the CLAUDE.md and AGENTS.md files you already wrote, feed every agent.
Can I use this?
You'll need macOS 13+ or Windows 10+ · Setup needed
Worth knowing Linux builds from the same source but the project says it is untested, so that platform is a build-it-yourself proposition. It is desktop software you install rather than something that runs inside your existing editor, so it is a change of workflow, not an add-on. Team synchronisation requires signing in and creating an organisation; without that, everything stays local. Apache 2.0. BuildTube has not run or verified this software.
Why it's here
- 2,590 people starred it on GitHub this week — 8,555 in total.
Numbers from the snapshot taken 1 October 2026; not refreshed since.
Behind it
See the code on GitHubBuildTube has not run or verified this project. Everything above is written from what the creator published.
Made this?