A Claude Code skill that wraps the OpenAI Codex CLI to inject decorrelated code reviews into review chains, helping catch blind spots that same-model reviewers miss. Launched with minimal external traction (2 HN points, 1 star) and no revenue model.

One insight: This is a narrow, read-only CLI wrapper with no moat, no monetization path, and virtually no adoption signal—a developer tool for an esoteric multi-model orchestration use case that would take 3 days to replicate but has near-zero wedge potential in a market where single-model AI review (Copilot, Claude) already dominates.