rag-blueprint

What problem does it solve?

This Skill eliminates the trial-and-error needed to stand up and maintain an NVIDIA RAG stack by automatically routing requests to the right deployment, configuration, troubleshooting, and teardown procedures.

Core Features & Use Cases

• Intent-based deployment & lifecycle management: Deploy, configure, debug, enable/disable, and shut down RAG components by interpreting user intent and selecting the correct operational reference flow.
• Cross-platform orchestration (Docker, Helm/Kubernetes, and library mode): Detects what’s running locally (containers/pods/processes) and applies changes in the correct mode and config location.
• Operational readiness, verification, and guided recovery: Performs environment analysis, health checks, platform detection, GPU/VRAM checks, and routes failures to troubleshooting playbooks without destructive actions unless explicitly requested.
• Feature configuration coverage for the full RAG pipeline: Handles VLM, guardrails, query rewriting/decomposition, ingestion types, search/retrieval tuning, model/profile changes, summarization, observability, multimodal querying, MCP integration, and migration planning.

Quick Start

Deploy the RAG stack and get it healthy locally using the command: run rag-blueprint to deploy RAG.