GitHub Work Habits Analyzer Measures AI Shift

This is a useful antidote to AI productivity theater: GitHub activity is imperfect, but longitudinal patterns can reveal whether AI is changing what developers actually ship.

• –GitHub reports that over 40% of pull requests began with an agent in September 2026, up from 17% in March, making before-and-after comparisons increasingly relevant. [GitHub Octoverse](https://github.blog/news-insights/octoverse/octoverse-commits-pushes-reach-new-highs-in-2026-as-developers-build-with-agents/)
• –MIT research found Copilot users shifted toward core coding and away from managerial tasks, suggesting useful metrics go beyond commit counts. [MIT IDE](https://ide.mit.edu/insights/beyond-productivity-genai-and-the-changing-nature-of-work/)
• –GitHub metadata cannot capture uncommitted exploration, review quality, private work, or AI usage without detectable signals.
• –Its strongest use case is personal benchmarking over time, not ranking developers or measuring raw productivity.