← Back to Glossary
Context Injection
Phase 03 · RAGtechnique

Formatting and inserting retrieved document chunks into the LLM prompt right before generating an answer.

Why it exists

Bridges the gap between static vector database storage and the active LLM context window.

How it works

Constructs templates like `Context: [Chunks]

Based ONLY on the context above, answer: `.

Where you'll use this