Evals / pdf

Organize pages

6 scenarios, 6 run by an agent. Each run shows its result, time and new tokens.

ScenarioCodexwithout deftCodexwith deft, skill installedCodexwith deft, skill as agents.mdClaude Opus 5.5without deftClaude Opus 5.5with deft, skill installedClaude Codewithout deftClaude Codewith deft
Merge two real formsCombine fw9.pdf and f1040.pdf, in that order, into tax-forms.pdf.pass14s · 6.4kpass15s · 4.3kpass10s · 9.2kpass37s · 12.2kpass7s · 9.3knot runnot run
Keep the first pages of a paperMake intro.pdf containing only the first 3 pages of paper.pdf.pass19s · 6.8kpass12s · 8.9kpass12s · 10.0kpass13s · 8.8kpass14s · 10.8knot runnot run
Remove pages from a formRemove pages 2 and 5 from fw9.pdf and save the result as w9-short.pdf.pass22s · 8.1kpass16s · 5.5kpass19s · 10.0kpass24s · 10.7kpass13s · 10.7knot runnot run
Split a 142-page document into chunksSplit pub17.pdf into separate PDFs of at most 50 pages each, in a folder called parts.pass42s · 10.8kpass19s · 8.4kpass13s · 10.7kpass33s · 11.2kpass20s · 10.7knot runnot run
Rotate only one pagePage 1 of fw9.pdf should be turned 90 degrees clockwise; leave the other pages alone. Save as turned.pdf.pass55s · 10.0kpass13s · 7.0kpass13s · 5.4kpass65s · 17.6kpass10s · 9.8knot runnot run
Trim page marginsCut 15mm off every side of the pages of paper.pdf and save the result as tight.pdf.pass42s · 10.1kpass23s · 7.6knot runpass26s · 11.6kpass12s · 9.9knot runnot run