rag-tutor

What problem does it solve?

This Skill prevents learners from receiving hallucinated or generic answers by always grounding responses in the Humanoid AI Studio curriculum, so learners get accurate, verifiable explanations and code examples.

Core Features & Use Cases

• Retrieval-first Grounding: Embeds queries and retrieves top curriculum chunks from Qdrant before generating any answer.
• Streaming, Grounded Responses: Streams Gemini-generated replies via SSE while citing module and lesson sources at the end.
• Robust Error Handling & Limits: Enforces per-user rate limits, retries Qdrant briefly on timeouts, and falls back to an alternate LLM if Gemini quota is exceeded.
• Use Case: A learner asks how to create a ROS 2 publisher and receives a grounded, lesson-cited example drawn directly from Module 1 content.

Quick Start

Ask Aria to explain how to create a ROS 2 publisher in Python using only the curriculum excerpts as context.