5 / 6
LLM evaluation

Test multilingual and emoji input

Medium70 pts~25 min
  • Unicode
  • Internationalisation
Practice app · Acme Support Assistant

A deterministic LLM-style support assistant with retrieval (RAG), JSON mode, safety policies and tool calls, exposed via UI and API.

BASE_URL
/api/practice
Console app
/lab/ai-testing-test-multilingual-and-emoji-input

Your starter code already declares BASE_URL — call the API relative to it.

Objective

Check that emoji and non-English text do not break the API and never produce invented facts.

Your task

  1. 1Send "📦 How long does shipping take? 🚚🙏" → assert 200, refused false and output contains “3–5 business days”.
  2. 2Send "¿Cuánto tarda el envío?" → assert 200 and that the answer either contains the documented shipping facts or says “I don't have that information”.
  3. 3Assert no response contains replacement characters (�).

Acceptance criteria

  • POST /ai/chat returns 200
  • At least 3 assertions pass

LLM evaluation · AI Testing · Prompt & input testing