Raula Gaikovina Kula

Final code diffs conceal catastrophic agent reasoning: mapping dense execution logs into classified, chronological action streams reveals precisely where models derail or burn tokens during repository navigation.

GenAI adoption in coding shifts the burden of maintenance from coding to extensive validation and dependency management, fundamentally altering developer workflows.

Newcomers beware: the odds of your "good first issue" pull request getting merged have plummeted nearly 20% in the last year.

Humans build, agents fix: AI-authored pull requests reveal a surprising division of labor in collaborative coding.

Catch duplicated effort and delayed reviews in large software ecosystems by surfacing semantically linked code changes directly within the code review interface.

AI coding agents mostly write beginner-level Python, suggesting current models may not be pushing the boundaries of code complexity.

Even in communities not primarily composed of developers, plugin ecosystems can develop and sustain themselves through recognizable engineering structures.