Skip to content

bamdra/Loom

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

6 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Bamdra Loom

Bamdra Loom

Tell Loom what you want — its AI team builds it. You don't need to write or read any code.

latest release license homepage

English · 简体中文


Bamdra Loom isn't a code editor or a chatbot. It's a small AI team that lives in your computer: you describe what you want; they argue about how to do it, write the code, review each other's work, and hand you the result. You never need to open code or learn jargon — you only say what you want and say yes or no.

What it does for you

🔬 Researchers and analysts

A folder full of PDFs, web clippings, raw spreadsheets — turn them into searchable little tools, dashboards, or topic-filtered web pages. Just say what shape you want; the agents wire it up.

🚀 People who want to build software but don't write code

That product idea you've been carrying around — a tool, a small game, a personal site, an app for friends — describe it in plain words. The agents ask back, decide on an approach, build it, run it, and hand you something you can open.

🛠️ Developers who want to stop hand-holding

Already shipping with AI? Drop the prompt-juggling. Say one sentence; the team runs the full analyze → plan → implement → review loop on its own. You only look at the result. Step in whenever you want.

Built by itself

Loom itself was built this way. The author started from one sentence — "I want a local desktop client for orchestrating multiple AI agents" — opened nothing but Loom, never wrote or read a line of generated code, and shipped this very app you're holding.

Why it works

Most AI coding tools optimize for one prompt → one patch. That breaks down on anything bigger than a single function. Loom takes a different stance:

  • Multiple agents, distinct roles — one drafts, another critiques, another implements. They keep each other honest.
  • Persistent context to files — plans, task lists, progress all land in .vibeflow/. Reopen the project tomorrow; any agent recovers the full picture.
  • A real loop with acceptanceintent → plan → tasks → implement → review → wrap up, with you standing at the start and the end.

Install

Download the installer for your platform from Releases:

Platform File
macOS (Apple Silicon) Bamdra Loom_<version>_aarch64.dmg
macOS (Intel) Bamdra Loom_<version>_x64.dmg
Linux x64 bamdra-loom_<version>_amd64.deb or .AppImage
Windows x64 Bamdra Loom_<version>_x64-setup.exe

Detailed install notes (including macOS quarantine handling) live in INSTALL.md.

After you install

  1. Open the app and pick a project directory (an empty one works fine)
  2. Type a sentence in the box at the bottom: "I want X."
  3. Look at the plan it shows you — accept or push back
  4. Watch it work; or come back when it's done
  5. Open the result; ask for tweaks if you want

The full walkthrough lives in docs/en/01-quickstart.md.

Updates

The app checks for updates at startup. When one is available, a download button appears in the top bar. The update channel is signed — only releases from this repository can install. If a download fails, a dialog will offer to retry or open the releases page directly.

Release channels

Two channels, distinguished by the tag form:

Channel Tag form Auto-update
Stable vX.Y.Z ✅ pushed to all users
Beta vX.Y.Z-beta.N ❌ manual download only

Stable users never receive beta builds — GitHub's "latest" automatically skips prereleases, and beta builds intentionally do not publish the updater manifest. To try a beta, grab the installer from the releases page directly.

Full conventions in RELEASE.md.

License

MIT © Bamdra

Links

About

Desktop AI orchestration workspace — turn intent into shipped code via multi-agent collaboration.

Topics

Resources

License

Stars

Watchers

Forks

Packages

 
 
 

Contributors