Agent-authored PRs hit 27.6% of GitHub pull requests

A DEV Community post dated October 3, 2026 examines how agent-generated pull requests are straining code review. The numbers it reports: over a fourteen-month span, agent-generated PRs went from under 1% of all GitHub PRs to 27.6%. The same piece cites Anthropic's 2026 Agentic Coding Trends Report, which gives a higher figure — 41% of all new code is now AI-generated, with workplace adoption of tools such as Claude Code up 6x in under a year.

The article's framing is blunt: the PR flood is here, and the review infrastructure is not. Its headline calls the problem the "Four Horsemen of Agent PRs and How to Stop Them," and the post goes on to discuss what the review step should do about that volume. The post treats the two data points as evidence that code generation has scaled faster than the process meant to check it. The excerpt available here carries the headline and the adoption and volume figures, but not the specific fixes the article proposes.

As of this report, that is where the story rests. One account has published these numbers, along with the argument that review processes are struggling to keep pace with agent-generated PRs. The material contains no later developments and no follow-up beyond the article itself.