agor

agor

Orchestrate multiple AI coding agents with your team

131 followers

Next-gen agent orchestration for AI coding. Multiplayer workspace for Claude Code, Codex, and Gemini.
agor gallery image
agor gallery image
agor gallery image
agor gallery image
Free
Launch Team
Intercom
Intercom
Startups get 90% off Intercom + 1 year of Fin AI Agent free
Promoted

What do you think? …

Maxime Beauchemin
**Hey, creator here, can give you a sense of what Agor actually is:** - **Real-time collaboration** with instant cursor sync (like Figma but snappier) - **Mobile site** for prompting agents on-the-go - **Built-in scheduler** for agent task management - **Full MCP support** + internal MCP server so agents are self-aware of the workspace - **Session forking/subsession spawning** - shown as session trees right on the canvas - **Shared dev environment management** across your team - **Information-dense but clean UI** that doesn't feel overwhelming Built the whole thing in a month using Agor itself with multiple agents collaborating. The meta experience of watching agents build the tool they're working in was... wild. Happy to answer questions!
Masum Parvej

@maxime_beauchemin How does Agor manage version control when multiple agents and humans are editing simultaneously?

Maxime Beauchemin

@masump Agor manages git worktree for you, meaning you add your repo, create one worktree per "feature", and start an agent on that worktree. Meaning big button top-right of the board creates a new worktree on one of your repos. Each "card" on your board is a worktree (think project/feature). Then in each on of these worktrees you can create AI sesssions, typically one at-a-time while coding, but you absolutely can have multiple agent on the same worktree if you want. You probably want to isolate their work if they are coding and make them aware of one another, like "can you commit all the work we've done so far, make sure to include the work of the other agent working on the frontend", or "NOTE: there's another agent working on this repo, only commit the work you've done". Also note that each new message in the AI session keeps track of the git sha at the time of the prompt. It's show a tag for each message like `31239a11` or `31239a11 (dirty)`. Generally want to keep it to "one actively coding agent per worktree", but in theory can have as many as you want, have Codex review claude code's work while its working, or whatever makes sense to support your workflow.

Maxime Beauchemin

@masump but yeah, the typical workflow is you create one worktree for each feature (each worktree, which is a card on the board is, or can be associated with a single issue, and ultimately PR), and have a bunch of cards on your board, each one with an agent actively working. I should have started with this, but here's the docs of the worktree feature in git Agor is using behind the scene -> https://git-scm.com/docs/git-worktree , allows you to have multiple branches checked out at the same time.

Sanskar Yadav

Congrats on the launch! The multiplayer workspace concept for AI coding is really unique.

How do you handle merge conflicts when multiple agents are working on the same codebase simultaneously?

Maxime Beauchemin

@sanskarix I explained in the other comment above, but yes, essentially Agor manages `git worktree` for you (a powerful native git feature allowing you to have multiple isolated branches checked out all at once). Each card on the board has a different/isolated branch checked out, think of it as different clones of the same repo.

Lilou Lane

Love the concept of tools that build themselves — feels like a glimpse into the future of creative collaboration. The clarity of your design choices really shines through. Congrats on the launch!