Summary
The Microsoft Developer Blog published a post testing whether AI coding agents pay attention to version numbers. The authors gave agents the same breaking upgrade twice, once labeled 2.0.0 and once labeled 1.4.3. According to the post, Claude Sonnet 5.5 checked the upgrade in all 10 runs under the 2.0.0 label, but in only two runs under the 1.4.3 label, where it often left the app broken. The post cites nearly 1,000 runs to show which agents still read version numbers and how a patch label affects their behavior.
AI-assisted summary based on the listed source.
What happened
We gave AI coding agents the same breaking upgrade twice. Labeled 2.0.0, Claude Sonnet 5.5 checked it in all 10 runs. Labeled 1.4.3, it checked in two and often left the app broken. Nearly 1,000 runs show which agents still read your version numbers, and what a patch label talks them out of. The post Do AI coding...
Why it matters
Teams that rely on semantic versioning to signal breaking changes may need to consider whether their AI coding agents read those labels. The reported results suggest that a patch-level label can make an agent less likely to check a breaking change, which could leave an application broken. Developers should watch whether other agents show similar patterns in the full results.
Signal Intelligence
Signal Strength 91%
Technical label SOURCE-BACKED
Public Interest 26
Category AI AT WORK
Reader Depth TECHNICAL
Signal Strength reflects source quality, relevance, freshness and evidence. Public Interest helps organize discovery; it is not proof of truth.
Public Interest components
Recognizable Entity Score 0
Practical Impact Score 28
Novelty Interest Score 48
Consequence Score 30
Curiosity Score 16
Shareability Score 42
Why this is here
VQV surfaced this signal because it is recent, relevant to AI Coding Tools, connected to Microsoft Developer Blog.
