claude-adversarial-reviewlisted

# Claude Adversarial Review Use this skill for high-risk changes, architecture decisions, reliability-sensitive code, security-sensitive code, migrations, rollback-sensitive changes, or when Codex may be overconfident. Run: ```bash node "${CODEX_PLUGIN_ROOT}/scripts/claude-companion.mjs" adversarial-review "$ARGUMENTS" ``` ## Natural-Language Claude Routing Codex should let the user ask for Claude adversarial review in normal language. Do not ask the user to write `--quality`, `--model`, or `--effort` unless troubleshooting the plugin itself. When converting the user's request to companion invocation: - Default to `--quality auto` for manual Claude review, plan, rescue, and multi-review unless the command documents a stricter default. - Use this skill when the user asks Claude to challenge assumptions, rollback paths, failure modes, architecture, hidden security risk, or removable complexity. - Use `--parallel` only when the user asks for independent adversarial lenses or parallel skeptical agents. - Use `--quality strong` for strict, deep, high-risk, security-sensitive, or release-sensitive adversarial review. - Use `--quality max` only when the user explicitly asks for the strongest local adversarial Claude review. - Before choosing foreground for an unforced review, run `recommend-execution-mode --json` or apply the same threshold: foreground only for tiny one-to-two-file work; use background for untracked directories, more than two files, more than roughly fifty cha