Native multiplayer coding for Mac. Work with your team and local agents, review patches, and land changes. Connect multiple provider accounts, track limits, and choose your theme. The latest development build adds existing Codex/Claude chat import and continuation, plus Codex image generation and comment-guided editing inside chat. Originals are kept. Provider credentials and agent execution stay on each Mac.
How did Astra change the scope or ambition of what you built?
Hunter
Astra changed cood from a fragmented set of prototypes into a product I could actually ship. It audited the native Mac app and older worktrees, helped unify the workspace, rebuilt multiplayer coordination and review, and fixed permission and account-isolation issues. That let me raise the goal from making several coding agents usable to making people and their existing local agents work together in one native workspace. Astra also stayed with the project through tests, signing, notarization, and the public alpha. I used other models for a few bounded QA checks; Astra led the engineering and release work.
All the UI/UX refactoring and beefy logic & architecture building was on Astra :)
Report
Hunter
📌
Hi Product Hunt, I’m Sergey, the maker of cood.
I built cood because my coding work was scattered across agent apps, prompts, branches, screenshots, and manual handoffs. cood puts people and the coding agents they already use into one native Mac workspace.
Since September 5, GPT-6 Astra has been the lead engineering agent on the project. It audited a fragmented codebase and older worktrees, helped unify the native workspace, rebuilt multiplayer coordination and review, fixed permission and account-isolation issues, and helped take the Mac app through testing, signing, notarization, and a public alpha release. I used other models for a few bounded QA checks; Astra led the build and release work.
The current release supports solo and multiplayer Rooms with Codex, Claude, Grok, and other installed agents. You can see live work, keep tasks, patches, and handoffs in one place, review changes, and land the result. Agent execution and provider credentials remain on each Mac.
I’d love to hear whether this makes multi-agent work easier to follow, and what would stop you from trying it on a real project.
Astra changed cood from a fragmented set of prototypes into a product I could actually ship. It audited the native Mac app and older worktrees, helped unify the workspace, rebuilt multiplayer coordination and review, and fixed permission and account-isolation issues. That let me raise the goal from making several coding agents usable to making people and their existing local agents work together in one native workspace. Astra also stayed with the project through tests, signing, notarization, and the public alpha. I used other models for a few bounded QA checks; Astra led the engineering and release work.
All the UI/UX refactoring and beefy logic & architecture building was on Astra :)