Sibling's proposed approach is a recipe for false positives. Consider Claude Code's new /security-review prompt or just use the prompt starting from "Objective

Sibling's proposed approach is a recipe for false positives. Consider Claude Code's new /security-review prompt or just use the prompt starting from "Objective" after giving the context (see the Git shell commands) to Claude Opus 4.1. https://raw.githubusercontent.com/anthropics/claude-code-security-review/refs/heads/main/.claude/commands/security-review.md https://raw.githubusercontent.com/anthropics/claude-code-sec... Definitely Opus 4.1 though, not lesser.