run “/claude-api prompt-audit” in Claude Code. this checks you skills, agent.md, Claude.md, prompts and removes anti-patterns that hobble frontier models.