Weave introduces AI Skills to score how engineers manage coding agents
The framework evaluates planning, verification, context, parallelization, collaboration and ownership from engineers’ conversations with agents.
Read the source ↗Weave’s AI Skills evaluates the decisions behind an AI-assisted code change, not just the finished patch. According to Weave’s breakdown, it looks at how engineers set direction, supply context, check results and retain responsibility for consequential decisions. Behaviors that aren’t needed or demonstrated in a conversation are excluded from the score rather than automatically counted against the engineer.