Multirun: A Multi-Agent Run Framework

Running a coding agent multiple times in isolated environments reveals model variability, enabling evaluation and informed branching decisions while reducing co