1 paper
Priyansh Trivedi, Olivier Schmitt
As autonomous coding agents see rapid adoption, their evaluation has primarily focused on task completion rates holding the target codebase fixed. This leaves a critical question u…