Lines of code are back as a metric in agentic coding

What the research found

Opinion piece arguing lines-of-code metrics returned with agentic coding, and high-performing teams now focus on delivery speed and reliability; cites CircleCI and DORA data showing AI-driven feature-branch gains with main-branch declines.

Claims and how well the evidence supports them

1. CircleCI 2026 report: median team feature-branch throughput rose 15% while main-branch throughput fell 7% 98%
2. DORA survey of nearly 5,000 professionals found higher AI adoption associated with both higher throughput and higher instability 99%
3. DORA described tokenmaxxing, ranking developers on raw token consumption 97%

Evidence

• cio.com https://cio.com/article/4236554/nobody-brags-about-lines-of-code-twice.html

Actionability rated low.