RA
RAG Architecture Review
🔵 Stable🕐 updated 2026-06-29
🔷 SkillSpec L3
pm-ai
Review an existing Retrieval-Augmented Generation system and find why it underperforms. Use when asked to review or audit a RAG pipeline, diagnose wrong/ungrounded answers from a 'chat with your docs' feature, or improve an already-built knowledge assistant. Produces a staged review — ingestion, chunking, retrieval, reranking, generation, evaluation — with prioritised findings, root causes, and concrete fixes.
What to give it
▸The current architecture — ingestion, chunking, embedding model, vector store, retrieval (top-k, hybrid?), reranking, and the generation prompt.
▸The symptoms — examples of bad answers (wrong, ungrounded, stale, refuses) with the expected answer.
▸The corpus — what's retrieved over, its size, structure, and update frequency.
▸Constraints — latency, cost, and per-tenant/permission isolation needs.
✅ The bar it holds itself to
Every skill in this library self-verifies — these are this skill's own quality checks, straight from its definition.
✓Every reported symptom is traced to a specific stage, not blamed on "the model"
✓Retrieval quality and answer quality are evaluated separately (or that gap is finding #1)
✓Findings are severity-ranked and the fix plan is ordered by impact, not by stage order
✓Hybrid retrieval and reranking are assessed for queries with exact terms/IDs
✓Grounding instruction and "I don't know" behaviour are checked in the generation stage
✓Per-tenant / permission isolation is verified in retrieval, not just the UI
⚠️ What it refuses to do
Do not recommend fine-tuning the model when the failure is in retrieval — fix what's retrieved first
Do not review only the generation prompt — most RAG quality is won or lost before the LLM sees anything
Do not present findings without severity and priority — a flat list doesn't tell the team what to do Monday
Do not assume the corpus is fine — stale or badly-structured source data caps every downstream stage
Do not skip the eval gap — without separated metrics, every fix is a guess
Install
npx pm-claude-skills add --agent claude # or codex · cursor · gemini · hermes
# or one-line MCP (every skill, any client):
claude mcp add pm-skills -- npx -y pm-claude-skills-mcpRelated skills
🔌 Embed this skill
Drop this on your blog, docs, or site — it renders a "Run this skill" card:
<div data-pm-skill="rag-architecture-review"></div>
<script src="https://mohitagw15856.github.io/pm-claude-skills/embed.js" async></script>💬 Discussion
RAG Architecture Review is one of 551 open-source professional AI agent skills — all SkillSpec L3.
Try them all in the browser · ⭐ Star on GitHub · Browse the full catalog