| name | rag-qa-engine |
| description | Build and evaluate a conversational QA layer over indexed enterprise knowledge with grounding, citation handling, and response quality checks. |
RAG QA Engine
Use this skill to shape a grounded question-answering layer over retrieved context.
Use for
- Prompt strategy for grounded answers
- Citation inclusion and fallback behavior
- Handling ambiguous or out-of-scope questions
- Basic evaluation criteria for faithfulness and relevance
- Runtime guardrails for unsafe or low-confidence responses
Do not use for
- Infrastructure provisioning
- Bulk indexing pipeline design
Output contract
Provide:
- Prompt and response policy
- Citation and confidence behavior
- Failure-mode handling rules
- Evaluation rubric and sample test prompts
- Short tuning backlog