About LilyCTM/claude-mentor-skills
Three drop-in skills for Claude Code (and any agentic coding session) that counter failure modes observed repeatedly in agentic coding sessions with strong models — especially Opus 5: defending conclusions after their reasons die, litigating against the user instead of helping, confidently inventing numbers, writing code on unverified assumptions about the layer beneath, tests that prove nothing, and changelogs that claim fixes for bugs the session itself introduced.