RAG Pipeline Vector DB

What this Claude Code skill does

Designs a traditional RAG pipeline organized around the Query-Retrieve-Augment-Generate (QRAG) flow and its single-LLM-call invariant, grounding answers in private documents via vector database retrieval with disciplined prompt augmentation and context injection. Use when building a RAG-powered chatbot or knowledge assistant over documents the base LLM was never trained on, or when answers drift b

RAG Pipeline Vector DB is a packaged skill for Claude Code: a SKILL.md plus references and a runnable test that drops into .claude/skills/, so your agent can run this workflow reliably instead of you re-explaining it each time. Install it via MCP, run it as a CLI, or fork it into your own.

Inside the SKILL.md

<h1>RAG Pipeline — Vector Database Retrieval (QRAG)</h1>
<h2>Why This Is Hard</h2>
<p>A base LLM has no knowledge of private or domain-specific information — internal
catalogs, proprietary documents, or any corpus that post-dates its training
cutoff. Without an external retrieval mechanism it cannot answer those queries
reliably. Three failure modes recur:</p>
<ol>
<li><strong>Hallucination on domain-specific queries</strong> — the model confabulates
   plausible but factually absent answers because the domain was never in its
   training data.</li>
<li><strong>Misuse of parametric memory</strong> — leaning on the model's weights to serve
   use-case-specific data that changes frequently guarantees stale answers.</li>
<li><strong>Context-limit overflow</strong> — naively stuffing the whole corpus into a prompt
   is cost-prohibitive and exceeds every production context window.</li>
</ol>
<p>The fix is not a bigger model or a bigger prompt. It is a disciplined
four-stage flow with exactly one LLM call.</p>
<hr />
<h2>Core Framework: The Query-Retrieve-Augment-Generate (QRAG) Flow</h2>
<p>QRAG is a four-stage sequential pipeline where the LLM is invoked <strong>exactly
once</strong>, solely to synthesize retrieved context into the final answer. That
single-LLM-call invariant is the defining property of <em>traditional</em> RAG and
…