[waleedlatif1]

Summary

• GET /api/auth/oauth2/authorize?providerId=…&workspaceId=…[&credentialId=…] picked the workspace (and, on reconnect, the credential) from the query alone and created a connect draft on the viewer's session. Session cookies are SameSite=Lax, so a cross-site page could navigate a signed-in user there and link their provider account as a credential in a workspace the attacker shares with them, or rebind an existing credential to the victim's account (CSRF on a state-changing GET)
• Reject that branch when Sec-Fetch-Site is cross-site, before any draft is created, using the existing isCrossSiteSessionRequest helper (same gate as workflow execute)
• draftId launches are unchanged: drafts are bound to the user who created them, and v2 API callers hand those authorization URLs to users from their own origin
• In-app (same-origin), www ↔ apex (same-site), and desktop/CLI/address-bar (none) launches still work. A server-side redirect such as /desktop/connect keeps the attacker's page as the initiator, so the browser still sends cross-site
• The proxy showed the login form to signed-in users for any /api/auth/oauth2/authorize callback, so /login?callbackUrl=<workspace launch> turned the launch into a same-origin request after sign-in. That exception now applies only to OAuth client requests (client_id present); a workspace launch callback sends a signed-in user to the app entry
• Shrinks the check-test-types baseline for the authorize route test (its GET calls now pass createRouteContext({}))

Type of Change

• Bug fix

Testing

• proxy.test.ts: signed-in /login with a workspace-launch callback redirects to the app entry (red before the change); an OAuth client callback still shows the form
• route.test.ts: cross-site workspace launch is refused with no draft or link call (red on the pre-fix code, confirmed by reverting the guard); a cross-site launch of the user's own draft still reaches the provider (the same-site/none header matrix is owned by same-origin.test.ts)
• bun run --cwd apps/sim test on authorize/route.test.ts + same-origin.test.ts + proxy.test.ts + app/(auth) suites (107 passed), bun run lint, Phase A generators (no drift); type-check, check:audits and the full suite run in 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)

🤖 Generated with Claude Code

[vercel]

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

Project	Deployment	Actions	Updated
docs	Ready	Preview	Nov 7, 2026 8:07pm UTC

[greptile-apps]

Confidence Score: 4/5

[Medium impact] The changes affect the bounded workspace OAuth account-linking and login callback flows.

The behavior changes appear safe, but the new test wrapper must use the required shared helper before merging.

Findings

1.  Test bypasses shared helper ▶

Summary

The PR blocks cross-site workspace OAuth launches before creating a connection draft.

• Workspace OAuth launches reject cross-site starts before creating a draft.

_{Reviews (3) · Last reviewed commit: "fix(oauth): keep workspace launches out ..." · Reviewed by Greptile}

[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}

[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 5 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}

[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 5 files

Confidence score: 5/5

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

_{Re-trigger cubic}