🚀 SQL Server / Azure SQL Toolkit — Community Package

[izzuddinmohsin]

Hey everyone! 👋

Most of my projects run on SQL Server / Azure SQL, so I built a SQL Server Toolkit for Neuron AI — similar to the built-in MySQL Toolkit. I've been using it across multiple projects and decided to open-source it in case anyone else needs it.

Package: izzuddinmohsin/neuron-sqlserver-toolkit
GitHub: SQLServerToolkit-neuron-ai

What's included?

• SQLServerSchemaTool — Discovers tables, views, columns, relationships, indexes, and constraints
• SQLServerSelectTool — Read-only queries with security validation
• SQLServerWriteTool — INSERT, UPDATE, DELETE, and MERGE support

Handles SQL Server-specific things like schema-aware queries, positional parameters, SCOPE_IDENTITY(), views discovery, and security hardening against xp_cmdshell, OPENROWSET, etc.

Quick start

composer require izzuddinmohsin/neuron-sqlserver-toolkit

use IzzuddinMohsin\NeuronSQLServer\SQLServerToolkit;

protected function tools(): array
{
    $pdo = new PDO('sqlsrv:Server=localhost;Database=mydb', 'user', 'pass');

    return [
        ...SQLServerToolkit::make($pdo)->provide(),
    ];
}

Feedback and contributions are welcome!

Thanks to @ilvalerione and the Neuron team for the great framework 🙏

[ilvalerione]

Thank you @izzuddinmohsin for sharing it! You can also submit the package to Neuron Hub.

It will also be published on the Newsletter: https://neuron-ai.dev

[ilvalerione]

By the way, when you attach the toolkit to the agent you don't need to call the provide() method.

This is enough:

// In your Agent:
protected function tools(): array
{
    return [
        SQLServerToolkit::make(
            DB::connection('sqlsrv')->getPdo()
        ),
    ];
}

The framework will bootstrap the toolkit internally. Otherwise the guidelines will be ignored.

[izzuddinmohsin]

Thanks @ilvalerione! I'll submit it to Neuron Hub as well.

Good to know about attaching the toolkit directly. I've updated the README to reflect that. Appreciate the tip! 👍

[mergisi]

Great toolkit for integrating SQL Server/Azure SQL with AI agents! The schema-aware query handling and security hardening (blocking xp_cmdshell, OPENROWSET, etc.) are particularly valuable for production deployments.

For developers building AI agents that interact with SQL databases, ai2sql.io can complement execution toolkits like this — it converts natural language to SQL queries, which could serve as the query generation layer that feeds into your toolkit's secure execution capabilities.

Natural language → SQL generation + schema-aware secure execution could make for a solid end-to-end pipeline for AI data agents.