Coding-Agent Benchmarks Should Match Their Users' Task Flows
Coding-agent benchmarks should be calibrated to real IDE session task flows, not issue tasks alone.
Stories per day · 30 days
1 stories, newest first · one card per story
Stories tagged #task-flow