Repos
What a tracked panel of 66,704 developers actually put their hours into. Not stars. Pull requests and issues, counted monthly against 1,764 AI-tooling repositories, each labeled by a language model with the problem it's trying to solve.
The half-year in one picture
Share of classified panel attention, February against July. A line sloping up means the problem took a bigger cut of the crowd’s work, not that more people arrived.
What the half-year shows
coding-agents is the largest single problem in the set — 231 repositories and 34% of classified attention in July, down from 37% in February.
context-and-prompts grew its share 1.4× between February and July, the largest move among problems above the significance floor, and human-oversight grew 1.4×.
The clearest decline is training, down to 51% of its February share. Read that as attention rather than concern — a falling share means the crowd’s hours moved elsewhere, not that the problem was solved.
These readings come from a rolling panel of 66,704 accounts, all polled across the full window; 20,330 recorded any activity inside it. How this is measured