agentsclimarketplace

Rag evaluation matrix

Skill mouadja02/skills/skills/agent-eval/rag-evaluation-matrix

A curated collection of agent skills for your AI agents - engineering craft, prompt engineering, design, growth marketing, ...

Install
npx -y skills add mouadja02/skills --skill rag-evaluation-matrix

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

2 things to look at

  • no licenseNo license file was found in the repository. Code published without one is not open source by default, so using it at work is a question for whoever answers licensing questions where you are.
  • 8 stars8 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.

What its author says it does

Copied from the file, not written here

Use when comparing basic, enhanced, GraphRAG, or agentic RAG designs, evaluating domain-specific RAG quality, tuning retrieval components, or deciding whether agentic RAG is worth its cost.

SKILL.md

3.0 KB, as published. Nobody here has run it

RAG Evaluation Matrix

Do not choose agentic RAG by fashion. Compare retrieval designs against domain questions, answerability, correctness, cost, latency, and failure reasons.

Use When

  • The user asks whether to use basic RAG, enhanced RAG, GraphRAG, or agentic RAG.
  • A RAG system works on demos but fails on domain-specific questions.
  • You need to compare embedding models, rerankers, chunking, query rewriting, or tool orchestration.
  • LLM-as-judge scores need alignment with human review.

Evaluation Flow

  1. Build a domain question set with answerable, unanswerable, ambiguous, multi-hop, and adversarial cases.
  2. For each candidate pipeline, log retrieved evidence, final answer, citations, latency, token usage, and cost.
  3. Score answer correctness and answerability separately.
  4. Add evidence quality checks: citation support, contradiction handling, source freshness, and missing-source diagnosis.
  5. Segment results by question type instead of reporting only one aggregate score.
  6. Inspect low-correctness failures and map them to retrieval, synthesis, tool orchestration, or corpus gaps.
  7. Pick the simplest design that meets quality, latency, and cost constraints.

Matrix

DimensionBasic RAGEnhanced RAGAgentic RAG
Best forStable FAQ and narrow corporaNoisy corpora, query mismatch, rerankingMulti-step, ambiguous, tool-rich tasks
Main riskWeak recall and unsupported answersPipeline complexityCost, latency, loops, tool misuse
Eval focusRetrieval recall and citation supportComponent ablationsTrajectory, action choice, stopping behavior
Ship gateCorrectness and answerability meet thresholdAblation proves each module helpsAgentic gains justify extra cost

Script

Use the helper to combine per-run JSON metrics into a decision table:

python skills/agent-eval/rag-evaluation-matrix/scripts/rag_eval_matrix.py results/*.json

Expected JSON fields: pipeline, question_type, correct, answerable_correct, latency_ms, cost_usd.

Common Mistakes

MistakeFix
Optimizing average score onlyBreak down by question type and domain
Judging unanswerable questions as wrong by defaultScore answerability separately
Skipping human calibrationSample judge disagreements and tune rubrics
Choosing agentic RAG without ablationCompare against enhanced RAG at equal budget
Ignoring failure reasonsClassify each miss before tuning

References

Keep looking

Skills are one crate of 328,083. Ordering is by how many stacks a row turns up in, so the top of any crate is what has actually been picked rather than what has the most stars.