A grounded answer is one the model takes from text you supplied, not from memory. Grounding is the main practical defence against invented facts: give the model the relevant documents, mark clearly where they begin and end, tell it to answer only from them, and let it say when they do not contain the answer.
Asking for citations, such as the passage or document behind each claim, makes answers checkable by a person or by code. These guides cover grounding from the beginner level: giving the model the right context, building a small retrieval pipeline with citations, quick checks for hallucinations, and structuring long-context prompts so the model finds and uses the relevant passage.
The library lists all guides with filters for topic, level, category and reading time.