https://github.com/xreedev/sql-orchestrator-agent-gemini

A minimal Python prototype that uses the Google Gemini API to generate, optimize, and run agentic pipelines that coordinate and run SQL queries and fetch data
https://github.com/xreedev/sql-orchestrator-agent-gemini

agent agent-based-model agent-framework gemini llm python rag

Last synced: 6 months ago
JSON representation

A minimal Python prototype that uses the Google Gemini API to generate, optimize, and run agentic pipelines that coordinate and run SQL queries and fetch data

• Host: GitHub
• URL: https://github.com/xreedev/sql-orchestrator-agent-gemini
• Owner: xreedev
• Created: 2025-07-27T16:27:58.000Z (7 months ago)
• Default Branch: master
• Last Pushed: 2025-07-27T16:42:09.000Z (7 months ago)
• Last Synced: 2025-07-27T18:43:30.246Z (7 months ago)
• Topics: agent, agent-based-model, agent-framework, gemini, llm, python, rag
• Language: Python
• Homepage:
• Size: 4.88 KB
• Stars: 0
• Watchers: 0
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: README.MD

Awesome Lists containing this project

README

          
# SQL Orchestrator Agent

A Python orchestrator that directs user prompts to specialized agents—SQLAgent for database queries and GenericAgent for general questions—then formats and presents results via a FinalAgent.

## Overview

* **OrchestratorAgent**: Analyzes user input and decides which agent to invoke.

* **SQLAgent**: Constructs SQL queries based on user questions and returns the query string.

* **GenericAgent**: Handles general knowledge or reasoning questions.

* **FinalAgent**: Merges raw results (from SQLAgent or GenericAgent) into a human-readable response.

* **DataBase**: Wrapper around an SQLite database providing table creation and data retrieval.

## Workflow

1. **Read Input**: Continuously prompt the user until they signal to exit.

2. **Route Prompt**: Use the OrchestratorAgent to determine if the request is a database query or a general question.

3. **Execute and Retrieve**:

   * If routed to SQLAgent, generate an SQL statement and fetch results from the DataBase.

   * If routed to GenericAgent, compute the answer directly.

4. **Finalize Response**: Pass the fetched data or computed answer to the FinalAgent to format a user-facing reply.

## Components

* **Agents**:

  * `OrchestratorAgent`: Decision-maker for routing prompts.

  * `SQLAgent`: SQL query generator with knowledge of your database schema.

  * `GenericAgent`: Handles non-SQL tasks or open-ended queries.

  * `FinalAgent`: Produces the final conversational response.

* **Database Wrapper**:

  * Automatically creates required tables (`subjects`, `teachers`, `marks`) on initialization.

  * Offers methods for inserting sample data and retrieving query results.

## Tech Stack

* **Python 3.x**: Core language for agent orchestration and database interaction.

* **SQLite**: Lightweight file-based database for storing `subjects`, `teachers`, and `marks`.

* **GeminiAgent (via `gemini_agent`)**: Base class for all AI agents (Orchestrator, SQL, Generic, Final).

* **SQLite3**: Python module for direct SQLite access.

## Basic Setup

* Ensure Python 3.x is installed and all dependencies are available.

* Place the orchestrator, agent, and database wrapper modules in your project directory.

* Initialize the `DataBase` with your database file (e.g., `school.db`).