The English version is not separately written; it renders the same measured data. The comparison:
| Item | Chinese page | English page | Match |
|---|---|---|---|
| Measured on | 2026-11-14 | 2026-11-14 | ✓ |
| Duration | 1 min 22 s | 82 seconds | ✓ |
| Success / failure | 8 / 2 | 8 succeeded / 2 failed | ✓ |
| Files processed | 10 | first 10 by filename | ✓ |
| Artifact size | 3,736 bytes | 3,736 bytes | ✓ |
| Source files | unchanged | no source PDF modified | ✓ |
| Checklist | 4 items | 4 items | ✓ |
What the English version keeps
- The full instruction verbatim, including the load-bearing constraints ('descriptions must come from the body', 'list unreadable files separately')
- All four checklist items, matching the Chinese page one for one
- Failure and boundaries: scans without a text layer cluster in the failure section; try ten files first when the count runs to thousands
The one place English needs extra words
The Chinese recipe asks to print the first ten for format approval before writing the full index. The English says the same, but an English reader may ask whether skipping that step is acceptable — it is. The step exists so a human sees the format; it is not a technical constraint.
This recipe really ran (8 succeeded, 2 failed), so its run log is not empty. The Chinese page carries four findings; the English page carries the same four in the other language, not a summary.