Multi-step jobs
3 scenarios, 3 run by an agent. Each run shows its result, time and new tokens.
| Scenario | Codexwithout deft | Codexwith deft, skill installed | Claude Opus 5.5without deft | Claude Opus 5.5with deft, skill installed | Claude Codewithout deft | Claude Codewith deft |
|---|---|---|---|---|---|---|
| Assemble an application from mixed filesPut together one application.pdf for a portal: letter.docx first, then the photo DSCN0010.jpg, then fw9.pdf. Number the pages at the bottom. The portal rejects anything over 1MB. | fail104s · 31.3k | pass36s · 10.2k | pass92s · 31.2k | pass28s · 15.3k | not run | not run |
| Send a scan to an accountantscan.pdf is going to our accountant. Make the text searchable, keep it under 1MB, and protect it with the password hunter2. Save it as for-accountant.pdf. | pass64s · 8.7k | pass41s · 9.6k | fail600s · 21.7k | pass50s · 18.7k | not run | not run |
| Collect invoice line itemsThe invoices folder has three PDF invoices. Put every line item from all of them into items.csv, with a single header row. | pass65s · 33.3k | pass28s · 8.4k | pass18s · 15.9k | pass23s · 12.6k | not run | not run |