[waleedlatif1]

Summary

• A workbench run stopped by its own timeout reports its output and files, but both run paths checked the call signal before building that report. When the provider's stop round trip outlasted the report reserve, the call deadline had already passed and the run returned a bare timeout with the printed output dropped
• The cancellation check moves into throwIfRunTimedOut: a deadline that passes while the stop settles keeps the report, and a cancellation at any point stays a cancellation
• One-shot sandboxes and runs that did not time out keep the same checks as before

Type of Change

• Bug fix

Testing

• New session-sandbox.test.ts case: a 10s workbench run whose stop settles 3s after its own timeout. Red on current staging (bare AbortError: timeout), green with the fix
• lib/execution/remote-sandbox/ suite: 22 files, 592 tests pass, including the existing cancellation-while-listing case
• biome clean; full type-check, check:audits and the full suite are left to CI

Checklist

• Code follows project style guidelines
• Self-reviewed my changes
• Tests added/updated and passing (new tests pass the test-audit authoring gate)
• No new warnings introduced
• I confirm that I have read and agree to the terms outlined in the Contributor License Agreement (CLA)

[vercel]

The latest updates on your projects. Learn more about Vercel for GitHub.

1 Skipped Deployment

Project	Deployment	Actions	Updated
docs	Skipped		Nov 3, 2026 6:38am UTC

[cubic-dev-ai]

No issues found across 2 files

Confidence score: 5/5

• Automated review surfaced no issues in the provided summaries.
• No files require special attention.

_{Turn on auto-fix | Re-trigger cubic}

[greptile-apps]

Confidence Score: 5/5

[Medium impact] This PR appears safe to merge.

Summary

Keeps a workbench run’s printed output when its timeout stop finishes after the call deadline.

• Timed-out workbench runs keep their output when stopping finishes late.

_{Reviews (1) · Last reviewed commit: "fix(sandbox): keep a workbench timeout r..." · Reviewed by Greptile}

[waleedlatif1]

Superseded by #9617, which landed the same change on staging first (the cancellation check moved into throwIfRunTimedOut via throwIfCancelled, so a passed deadline keeps the timeout report). This PR's regression test, a 10s workbench run whose stop settles 3s after its own timeout, passes on current staging.