Model Key Policy
Madeenan does not ask customers to save third-party LLM provider keys. Production apps should retrieve sources from Madeenan, then call their own model provider from their own backend.
Recommended Flow
const sources = await fetch("https://api.madeenan.com/v1/search?q=patience&indexes=quran,hadith,tafsir,dua", {
headers: {
"X-Madeenan-API-Key": process.env.MADEENAN_API_KEY
}
}).then((res) => res.json());
// Next, call your own LLM provider from your backend.
// Put sources.results into the model context and render Madeenan citations
// beside the generated answer.
What Madeenan Stores
- Your Madeenan app, subscription, API key hash, and usage records.
- Islamic source retrieval results and citation metadata needed for responses.
- No customer OpenAI, Anthropic, Gemini, or other model provider API keys.
Security note
Hosted chat uses Madeenan's own configured Gemini Flash-Lite key and is quota limited. For production generation, use the Search API and keep your model credentials in your own infrastructure.