Rag Answer

GPTClaudeDeepSeek··▲ 460 copies·updated 2026-07-14

rag-answer-2.prompt

# Grounded RAG Answer

A system prompt for the generation step of a RAG pipeline. It answers **only** from the retrieved context, cites every claim, and says "I don't know" rather than filling gaps with model memory. This is the guardrail that keeps a retrieval system honest.

**When to use:** the final LLM call in any retrieve-then-generate flow — a knowledge-base assistant, doc Q&A, support bot, research tool. Put it in the **system** slot; pass the retrieved chunks and the question as the user message. Implements the generation half of rule [`50-rag-system`](../claude/rules/50-rag-system.md) (keep ingestion separate from generation); see also [`../foundations/`](../foundations/).

when to use it

Community prompt sourced from the open-source GitHub repo aiagentwithdhruv/ai-dev-stack (MIT). A "Rag Answer" style prompt — adapt the placeholders and specifics to your task. Imported as-is and not independently retested here, so check the output before relying on it.