domains / families / data / rag-evaluation-harness

rag-evaluation-harness

separate

Evaluate retrieval and citation behavior for RAG pipelines from deterministic JSONL fixtures. Use when an agent needs offline Recall@K, reciprocal rank, context precision, citation coverage, citation validity, diagnostics, Markdown/JSON reports, or threshold-gated evaluation in CI.

When to use

• offline Recall@K
• reciprocal rank
• context precision
• citation coverage
• citation validity
• diagnostics
• Markdown/JSON reports
• threshold-gated evaluation in CI

When not to use

• semantic answer quality
• entailment
• attribution correctness
• groundedness

Skill page

Source