01Active prevention of vague language like 'should' or 'seems to'
02Red-green cycle validation for bug fixes and regressions
03Real-time exit code and failure count analysis
04Independent verification of delegated agent task reports
05Mandatory command execution before any success claims
060 GitHub stars