claude-reviewlisted

# Claude Review Use this skill when Codex needs an independent Claude Code review before shipping. Run: ```bash node "${CODEX_PLUGIN_ROOT}/scripts/claude-companion.mjs" review "$ARGUMENTS" ``` ## Natural-Language Claude Routing Codex should let the user ask for Claude 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. - Keep strict-only, focused, or "double check this" requests in `claude-review`; do not fan out unless the user asks for multiple roles, multiple perspectives, subagents, or a named team. - Use `--quality strong` for "deep", "strict", "advanced", "high-confidence", "strong Claude", or "harder local Claude" review when the user does not name a concrete model. - Use `--quality max` only when the user explicitly asks for the strongest local Claude review or max local effort. - If the user names a concrete Claude model or effort, pass it as explicit argv tokens outside quoted `$ARGUMENTS`. - Use `--background` for broad diffs, unclear scope, or long reviews instead of blocking the main Codex turn. - 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 untr