# GitHub Octoverse reports record code activity and rising agent use > GitHub reports 17.9 billion pushes in its 2026 reporting year, but higher commit counts do not establish productivity gains from agents. Canonical URL: https://www.devobs.io/news/news-event-8afc6331e258c564363b70a7/ By: Jonah Reed Published: 2026-10-28T22:22:21.648Z Updated: 2026-10-28T22:22:21.648Z Event date: 2026-10-28 Section: Developer tools GitHub’s 2026 Octoverse report, published October 28, reports 17.9 billion code pushes during its reporting year, October 1, 2025, through September 30, 2026. That exceeds its recalculated 2024 and 2025 totals combined. Commits across public and private repositories reached 34.3 billion, more than three times the level two years earlier, according to [GitHub’s report](https://github.blog/news-insights/octoverse/octoverse-commits-pushes-reach-new-highs-in-2026-as-developers-build-with-agents/). ## Agent use and automation Among repositories created in the reporting year, those containing an agent configuration file averaged eight times as many commits as those without one. GitHub also says more than 40% of pull requests on the platform started with an agent’s first draft in September 2026, up from 17% in March. Automation activity was substantial too: GitHub Actions averaged 1,440 job executions per second, equivalent to about 124 million daily. These are job counts, not execution minutes. All three measures come from [GitHub’s findings and methodology](https://github.blog/news-insights/octoverse/octoverse-commits-pushes-reach-new-highs-in-2026-as-developers-build-with-agents/). ## Compare like with like The reporting window matters. The annual push total ends September 30—not on the report’s October 28 publication date—while the agent-started pull-request shares compare March and September snapshots. GitHub says it recalculated the 2024 and 2025 figures using the same definitions and reporting window as the 2026 report. Earlier reports covered public activity and used a September 1–August 31 window. Their originally published totals therefore should not be substituted into these comparisons. Unless otherwise noted, the 2026 metrics include public and private repositories. Pushes and commits are also different units: one push can contain multiple commits. These scope and counting distinctions are set out in [the report’s methodology](https://github.blog/news-insights/octoverse/octoverse-commits-pushes-reach-new-highs-in-2026-as-developers-build-with-agents/). ## Measure outcomes beyond activity The eightfold commit difference is an association, not evidence of an eightfold productivity gain. Grouping repositories by the presence of a configuration file does not, by itself, establish which commits an agent produced or whether agents caused the gap. Repository size, workload and commit practices are possible alternative explanations. For engineering teams, the useful next step is to evaluate outcomes alongside activity. Pair agent adoption with pull-request review latency, CI consumption, escaped defects and rollback rates. Ask whether additional activity produces useful changes without increasing quality problems or operational costs. Commit counts alone cannot answer that question. ## Source references -