# Android RAG Local

> Implement local Retrieval-Augmented Generation using Room and Vector embeddings.

- Skill: `prasad-vennam/android-rag-local` (Agent Skill)
- Install (CLI): `npx skillmds@latest add prasad-vennam/android-rag-local`
- Raw SKILL.md: https://api.skillmd.com/api/skills/prasad-vennam/android-rag-local/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: AI & ML
- Author: prasad-vennam (https://skillmd.com/u/prasad-vennam)
- Updated: 2026-09-22
- Page: https://skillmd.com/skills/prasad-vennam/android-rag-local

---


# Android Local RAG & Vector Databases 🔎

Retrieval-Augmented Generation (RAG) on Edge devices provides privacy-first, low-latency contextual intelligence. Instead of using cloud Pinecone/Weaviate, this skill enforces local vector embeddings via Room or SQLite.

## ⚡ When to Use
- When tasked with "RAG", "semantic search", or "vector embeddings".
- Building highly private AI features.

## 🏗️ Core Rules / Pillars

### 1. Vector Search in Room
- **Pattern**: Standard Room does not have native vector similarity out of the box in stable. Use L2 distance extensions or `sqlite-vss`.
- **Implementation**: Avoid pulling unmaintained third-party vector DBs. Use SQLite virtual tables with FTS or simple cosine similarity functions if embeddings are small.

## 🚧 Critical Anti-Hallucination Guards
- **Trap**: Do NOT hallucinate that Room natively supports `SELECT * FROM tbl ORDER BY VECTOR_DISTANCE()` securely out of the box without specific extensions.