Three AI agents tracked down fifteen rendering bugs, and a run over 28,932 test pages showed nothing else had broken
Fifteen pages came out of our converter looking wrong. Three agents went looking for the reasons, one agent applied the fixes, and a full test run checked that nothing else broke. In every case the answer was a value Chrome had already worked out and we had never asked it for. This write-up follows the order the work happened in: the situation, the task, what was done, and what came of it.
| Outcome | |
|---|---|
| Target pages passing | 12 / 15 |
| Corpus passing | 27,142, up from 27,087 |