Effect of Commit-Message Quality on Agent Repair Decisions

Determine how the quality of commit messages generated during commit untangling affects coding agents’ ability to identify relevant changes, preserve required behavior, and avoid introducing new regressions during software repair.

Background

RucTangle supplies coding agents with untangled commit histories and commit messages when they repair regressions. The paper’s trajectory analysis shows that agents use commit messages to infer why changes were made and which behavior should be preserved, but misleading messages can cause agents to question or abandon a correct repair. Because the experiments did not independently control message quality, the paper identifies the causal effect of message quality on subsequent repair decisions as unresolved.

References

We did not independently control message quality in our experiments, leaving its effect on repair as an open research question.

— Runnable Commit Untangling for Coding Agents  (2610.11593 - Jiang et al., 8 Oct 2026) in Section 6, “Implications and Research Opportunities,” paragraph “Evaluating commit messages through subsequent decisions”

Whether untangled histories also help agents perform code review and refactoring, or help human developers review and integrate changes, remains to be evaluated.

— Runnable Commit Untangling for Coding Agents  (2610.11593 - Jiang et al., 8 Oct 2026) in Threats to External Validity, Section 7