claude-multi-reviewlisted

# Claude Multi Review Use this skill when Codex needs several independent Claude review passes with different role directives. This is an opt-in fan-out review workflow, not automatic repair. Foreground role reviewers run in parallel by default. Run: ```bash node "${CODEX_PLUGIN_ROOT}/scripts/claude-companion.mjs" multi-review "$ARGUMENTS" ``` ## Natural-Language Claude Routing Codex should let the user ask for Claude multi-role 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 for multiple perspectives, named review roles, multi-agent review, role fan-out, SDK subagents, or a named reviewer team. - Treat "strict" alone as review strength, not role fan-out. Strict-only single review belongs to `claude-review`. - Use `--quality strong` for high-risk, security, release, migration, or adversarial multi-role review when the user does not name a concrete model. - Use `--quality max` only when the user explicitly asks for the strongest local multi-role Claude review. - Use `--agent-team sdk-subagents --backend sdk` only when the user explicitly asks for native Claude SDK subagents or native subagent orchestration. - Use `--background` for more than three roles, la