Author

Pranav Verma

Firebase RAG CLI (Firestore Natural Language Query Engine)

A Retrieval-Augmented Generation (RAG) system that allows you to query Firebase Firestore using natural language prompts, powered by Llama 3 via Ollama.

This tool converts plain English questions into structured Firestore queries and returns results directly from your database.

---

Features

• Natural language querying for Firestore
• Schema-aware query generation (required)
• Powered by Llama 3 via Ollama
• CLI-based initialization
• Firebase Admin SDK integration
• Fully local execution (no cloud dependency for query processing)
• Private key remains local to your machine during execution
• Simple setup and execution

---

Security & Privacy

This tool is designed with a local-first architecture:

• All query processing happens locally on your machine
• Llama 3 runs locally via Ollama
• Firebase service account key (firebase-key.json) is used only locally by Firebase Admin SDK
• The private key is never sent to any external server by this library
• Users provide their own credentials, and all database access happens from their local environment
• No user queries, schema data, or Firestore results are transmitted to third-party services by this tool

Note: The Firebase private key is used locally to authenticate requests with Firebase Admin SDK. It is not exposed to the internet by this library.

---

Prerequisites

1. Install Ollama and Llama 3

pip install ollama
ollama pull llama3

Make sure Ollama is installed and running on your system.

---

2. Firebase Setup

You must have a Firebase project and a service account key file.

Download your firebase-key.json from Firebase Console.

---

Installation

pip install pranavfirebase-rag

---

Initialization

After installing the package, run:

my-library init

This command will generate the following files in your project directory:

schema.json
firebase-key.json
rag.py

---

Configuration

1. schema.json (Required)

You must define your Firestore schema in this file.

Example:

{
  "users": {
    "Age": "int",
    "Name": "string",
    "Department": "string",
    "Salary": "int"
  }
}

This schema is mandatory and used for query parsing and structured retrieval.

---

2. firebase-key.json

Paste your Firebase service account credentials into this file.

Important:

• The private key is stored locally on your machine
• It is only used by Firebase Admin SDK for authentication
• This library does not transmit it anywhere

Example structure:

{
  "type": "service_account",
  "project_id": "your-project-id",
  "private_key": "-----BEGIN PRIVATE KEY-----\n...\n-----END PRIVATE KEY-----\n",
  "client_email": "your-client-email"
}

---

3. rag.py

This file is the main chatbot entry point.

Important:

• Replace the default collection name (e.g. employees) with your Firestore collection name
• Do not modify internal logic unless required

---

Usage

Run the chatbot:

python rag.py

You will enter an interactive terminal where you can query your database.

---

Example Queries

• Show all users above 25
• List employees in AI department
• Get users with salary greater than 100000
• Find all names in the users collection
• Show users younger than 30

---

How It Works

User Input → Llama 3 (Ollama) → Schema Parser → Query Builder → Firestore (Firebase Admin SDK) → Response Output

---

Architecture

User Input → Llama 3 (Ollama) → Schema Parser → Query Builder → Firestore → Response Output

---

Notes

• Schema definition is required (not optional)
• Firebase credentials must be valid
• Ollama + Llama 3 must be installed and running before execution
• Collection name must be correctly set in rag.py

---

Requirements

• Python 3.8+
• Firebase Admin SDK
• Ollama
• Llama 3 model

---