Coding agents skip documentation in most PRs, and undocumented changes are later fixed more often. Add doc-update checks to AI agentAn AI system that doesn't just answer once but works toward a goal in a loop — taking actions, reading the results, and deciding what to do next.Full definition → PR workflows.
Key takeaways · AI-distilled
Across 29,669 AI-generated pull requests in 2,585 repositories, the authors report that 63.2% changed code without updating any documentation file, rising from 55.2% for features to 77.5% for refactors.
Undocumented code files were followed by a fix to the same file 8.6-11.7% more often over 30-90 days. A naive PR-level comparison shows the reverse, because documented PRs are 3.9x larger; controlling for size flips the result.
Documentation debt compounds, per the abstract: within 90 days, files changed without documentation were later re-documented at about half the rate of documented ones (24% vs 48%).
Terms in this piece · Glossary
AI agent — An AI system that doesn't just answer once but works toward a goal in a loop — taking actions, reading the results, and deciding what to do next.
articleNot All Agents Are Equal: Code Quality and Post-Merge Maintenance Across Five Autonomous Coding Agents in the WildObada Kraishan
articleWho Finishes the Job? A Study of Follow-Up Fixes and Commit Authorship on AI Coding Agent Pull RequestsWannita Takerngsaksiri, Nhat Duong, Scott Barnett
articleA Few Pages of Markdown: Committed AI Configuration and Lower Quality Cost after Coding-Agent AdoptionYegor Denisov-Blanch, Shyam Agarwal, Pavel Azaletskiy, Hao He, Rylan Schaeffer, Brando Miranda, Bogdan Vasilescu, Sanmi Koyejo