Factory's Droid Shield 2.0 adds two learned checks around its existing scanner for access credentials in code. One examines suspicious lines that the scanner missed. The other revisits blocked lines to identify false alarms, with detected credential values masked so its judgment depends on surrounding code. These address different problems: overlooked secrets and unnecessary interruptions.
The research-preview report evaluates the models on public benchmarkA standard public test set for comparing AI models — the shared scoreboards behind every "model X beats model Y" claim.Full definition → repositories held out from training. It also acknowledges that private repositories may contain different patterns. The contribution is a context windowThe maximum amount of text a model can consider at once — its working memory for the current conversation or task.Full definition →-sensitive layer around scanning, with an explicit tradeoff between catching secrets and raising false alarms.
Terms in this piece · Glossary
benchmark — A standard public test set for comparing AI models — the shared scoreboards behind every "model X beats model Y" claim.
context window — The maximum amount of text a model can consider at once — its working memory for the current conversation or task.
