You are integrating Claude Code into your Continuous Integration/Continuous Deployment (CI/CD) pipeline. The system runs automated code reviews, generates test cases, and provides feedback on pull requests. You need to design prompts that provide actionable feedback and minimize false positives.

Your automated review generates many findings per PR, but developer feedback shows roughly half are dismissed as "not worth addressing." Analysis reveals dismissed findings are often technically accurate but involve minor style preferences or patterns acceptable in your codebase.

Before adding infrastructure complexity, what prompt design change would most effectively reduce dismissals while maintaining detection of genuine issues?