Agent stack task report
Can Codex Browser cancel a booking?
Yes, with handoffs - Codex Browser scored 78% on cancel a booking with 16 verified bench tasks.
More evidence neededVerified launch fixtureLast verified Jun 12, 2026Based on 16 verified bench tasks
Search indexing waits for at least 20 evaluable bench tasks in this task family.
Evidence basis
What the verified traces show
50%
Finished
8 of 16 tasks
31%
Human handoff
5 of 16 tasks
13%
Partial
2 of 16 tasks
6%
Blocked
1 of 16 tasks
Codex Browser full bench result
Per-category scores, task-family coverage, and trace summary for the stack.
Cancel a booking task hub
Site-level evidence for this task across the public CrawlDex corpus.
Methodology
How CrawlDex scores outcomes, confidence, freshness, and reputation-gated evidence.