[waleedlatif1]

Summary

• Workbench runs with a budget of 20 s or less give the command the whole budget, so the run's own deadline could fire before E2B's timeout reply. The run then threw a bare timeout and dropped the output it had already streamed. The E2B session adapter now treats a timeout-reason abort like E2B's own timeout and returns the streamed output with timedOut: true. throwIfRunTimedOut lets only the run's own deadline through (throwIfCancelled) and runs before the post-run abort check, so the caller gets the SandboxExecutionTimeoutError progress report. A Stop is still a cancellation, and one-shot sandboxes still get the plain timeout. This fix works alongside the report reserve from fix(release): report short workbench timeouts, gate Sim links in Chat settings, scope the Chat plan modal #9616: that reserve usually lets E2B's timeout answer first, and this change covers the case where the call's deadline still fires first. It takes no additional budget.
• Daytona is unchanged because it has no workbench (session) support.
• The nullable workspace.project_id dev-push bootstrap switch now also requires SIM_DEV_DB_PUSH=1, the same way hnswIndexOptions does. A stray SIM_DB_PUSH_PROJECT_BOOTSTRAP can no longer make a non-dev push or db:generate drop NOT NULL.

Type of Change

• Bug fix

Testing

• Each new regression test fails when its guard is reverted:
  • session-sandbox.test.ts: the deadline fires before the provider replies, for both code and shell, and the run reports the streamed output; it still throws a bare AbortError: timeout on staging with fix(release): report short workbench timeouts, gate Sim links in Chat settings, scope the Chat plan modal #9616. A Stop stays a cancellation on a workbench, and also on a provider without workbenches.
  • e2b-session.test.ts: a deadline abort returns the streamed stdout for both code and command. A Stop of code still rejects. A Stop of a command is never reported as a timeout. A one-shot sandbox's deadline still rejects.
  • dev-push.test.ts: with only the bootstrap flag set, project_id stays NOT NULL; with both flags it is nullable.
• packages/db suite; remote-sandbox session, E2B, conformance and session-files tests; execute-request.test.ts
• drizzle-kit generate produces no migration
• lint, check:comment-hygiene, check:test-patterns, check:utils, check:explicit-any. CI runs the full gate.

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 6:37am UTC

[cubic-dev-ai]

No issues found across 6 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] The PR appears safe to merge; no actionable issues were found.

Summary

Preserves streamed output when a workbench run's deadline fires before E2B replies.

• Short E2B workbench runs keep streamed output when their deadline fires.
• The Project bootstrap makes workspace.project_id nullable only during development pushes.

_{Reviews (2) · Last reviewed commit: "fix(sandbox,db): keep a short workbench ..." · Reviewed by Greptile}

[waleedlatif1]

@greptile

[waleedlatif1]

@cubic-dev-ai review this PR

[cubic-dev-ai]

@cubic-dev-ai review this PR

@waleedlatif1 I have started the AI code review. It will take a few minutes to complete.

[cubic-dev-ai]

No issues found across 6 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}