Repository navigation
feat(sandbox,ca): leave experimental, one create and ssh output for sandboxes and cloud agents - #1371
Merged
Merged
Conversation
Base automatically changed from
fix/ca-idle-sleep-copy-and-template-digest
to
master
November 17, 2026 19:23
Sandboxes and cloud agents are leaving experimental, so `railway sandbox`, `railway code` and the `railway ca` launcher no longer print the "experimental and APIs may change" banner. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Cloud agents no longer sit behind the CLOUD_AGENTS flag, so `railway ca`, `railway code`, `ca desktop` and the Codex/OpenCode clients stop asking for the account's feature flags before launching. This removes the access module, the CloudAgentAccess query, the flag_missing telemetry event and the "Requires Cloud Agents access" lines in help and docs. Older CLIs that still run the preflight keep working because the backend reports CLOUD_AGENTS for every user. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
A cloud agent is a sandbox with agent tooling on top, so `railway sandbox create` and `railway ca create` now print the same block from one renderer (commands/output/machine.rs): a "Created sandbox|cloud agent <label>" header, aligned Status, Region, Idle and Domains fields, then one Connect line. The agent adds a single dim line about installing a coding agent. `railway sandbox create` (and `fork`) now waits until the sandbox runs, with the SDK's 5 minute readiness timeout, and fails clearly on FAILED or a destroyed sandbox. `--no-wait` keeps the old return-while-CREATING behavior and prints the same block with the current status. JSON is additive: sandbox output gains `name` (null; sandboxes have none) and `projectId`; cloud agent output gains `region` and `domains`, read with a new CloudAgentPlacement query. `region` is added to CloudAgent in schema.json. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
`railway sandbox ssh` and `railway ca ssh` now show a short "Connecting to sandbox|cloud agent <label>" spinner (cloud agents: "Waking" when asleep) that clears before ssh takes the terminal, and one line after an interactive session: Disconnected from <kind> <label>. <what happens next>. <command to act now>. Sandboxes say when they are destroyed (or that they never idle out); cloud agents keep the idle-sleep wording and the old-agent split from ca::sleeps_when_idle. One-off commands (`ssh -- <cmd>`) print nothing extra in either, so `ca ssh -- <cmd>` no longer prints the disconnect note. Exit codes are unchanged. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
`railway sandbox --help` ended with "Requires Sandboxes access.", but sandboxes have no feature flag and Backboard allows them on Free, Trial and paid plans. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
codyde
force-pushed
the
feat/sandbox-agent-consistent-output
branch
from
November 17, 2026 19:28
13e73cd to
1175cae
Compare
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Sandboxes and cloud agents leave experimental. This drops the warning banners and gives both the same create and ssh output, since a cloud agent is a sandbox with agent tooling on top.
Changes
railway sandbox,railway codeand therailway calauncher.CLOUD_AGENTSPriority Boarding preflight (access.rs, theCloudAgentAccessquery, theflag_missingevent, "Requires Cloud Agents access" in help and docs).sandbox createandca createprint one block from a shared renderer (commands/output/machine.rs): header, Status, Region, Idle, Domains, then a Connect line. The agent adds one dim hint line.sandbox createandsandbox forkwait until running (5 min, the SDK'sREADINESS_TIMEOUT_MS) and fail clearly on FAILED or a destroyed sandbox.--no-waitreturns at once and prints the same block with the current status.name(null) andprojectId; cloud agents gainregionanddomains(newCloudAgentPlacementquery,regionadded toCloudAgentin schema.json).sandbox sshandca sshshow a "Connecting to ..." spinner ("Waking cloud agent ..." when asleep) and oneDisconnected from ...line after an interactive session. One-offssh -- <cmd>prints nothing extra in either, soca ssh -- <cmd>drops its disconnect note.--jsonkeeps it.railway sandbox --help: sandboxes have no feature flag and are allowed on every plan.Before / after
railway sandbox create --private-network --domain web:3000railway ca create my-boxDisconnect lines
Notes
railway sandbox createnow waits until running; use --no-wait for the old behavior. The docs say create returns while CREATING (railwayapp/docs #1453) and need a one-line update once this releases.railway ca ssh <name> --sessionrather than fix(ca,sandbox): idle-sleep copy, retire --base-image-digest #1369'srailway ca ssh <name>: a plainca sshopens a login shell and installs nothing;--sessionstarts the default harness.railway code, the TUI and herdr keep their own create and "Session ended" messages, andca wakestill says "agent".🤖 Generated with Claude Code