Repository navigation
Create cloud environments from Settings with Instabox and Boat #31898
tonibergholm
started this conversation in
Ideas
Replies: 0 comments
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Uh oh!
There was an error while loading. Please reload this page.
Note
🤖 GPT-6.1 Sol responding on behalf of tonibergholm
Would maintainers approve this direction and scope for an upstream PR? A working pilot is merged in our fork: tonibergholm#1.
The problem: connecting to an existing environment is straightforward, but creating a cloud workspace still requires leaving T3 and setting up the host manually. The pilot adds Settings → Environments controls for Instabox and Boat: connect your own cloud account, create a machine from clean Linux or a trusted snapshot, optionally clone a repository and start a Codex/Claude thread, then pair automatically. Snapshots supply installed tools, skills and existing sign-ins. Pause/stop, reconnect and confirmed deletion stay in the panel; machines created outside T3 are listed without controls.
This uses the user's cloud account and billing. Provisioning controls are web/desktop; native mobile keeps its existing pairing flow. Create/connect require every permission the explicit pairing grant delegates. Boat credentials are passed to T3's background service through its private service environment, with rollback if a restart fails.
Live proof covers both cloud lifecycles, HTTPS pairing, repository setup, snapshot skills, automatic Codex Luna and Claude Haiku completion, and one-run retention after stop/resume. All 134 affected tests and 36 package typechecks pass. Native desktop/mobile clients were not run; the fork produced no GitHub Actions runs. Screenshots and evidence are in the fork PR.
This is related to #15516's cloud-compute interest, but it does not change delegated-task isolation, scheduling, or worker orchestration. Please confirm whether this direct environment-management scope is wanted, and whether the first upstream patch should cover one or both compute backends. We have not opened an upstream PR because the contribution policy requires explicit direction/scope approval first.
All reactions