The AI Slop Detector, a tool for identifying code that looks complete but lacks substance, released version 2.6.2. This update directly addresses a specific failure mode: projects with green CI pipelines that still feel hollow. The core issue is code that passes tests but doesn't hit real dependencies, creating a dangerous illusion of stability.

The release was prompted by feedback from OnlineProxy, who highlighted a repo where CI was green but zero integration tests touched actual systems. This isn't a minor quibble; it's a real production risk. The developer treated the comment as a bug report, shipping v2.6.2 as a targeted patch to make this gap measurable and reportable.

Version 2.6.2 now explicitly splits test analysis into unit and integration categories, using path, filename, and marker detection. More importantly, it ties claims to evidence: calling code production-ready now requires both unit and integration tests. The goal is clearer reports that warn when strong claims lack corresponding real-world verification, moving beyond mere test counts.