1. Multi-adapter demo (same task -> every available adapter, compare)

npm run demo

Dispatches one fixed task to all available adapters in parallel and prints a comparison table (provider | status | output | costUsd | ms) plus per-adapter detail. Proves every adapter returns the same Result shape regardless of runtime. It then coordinates the results across three Judge strategies (consensus, llm-judge, adversarial) and prints one verdict line per strategy (winner + agreement + rationale). The LLM-backed judges run only when a judge model is configured (OPENROUTER_API_KEY).