"DTUs comin' in hot!"
An AI agent skill that monitors SQL Server performance and delivers actionable insights when your database metrics spike.
The Problem
When SQL Server performance degrades, you usually find out via alarms after the damage is done. The metrics live in separate monitoring tools, and getting actionable insights requires manual investigation. By the time you've pulled the data, your users have already felt the pain.
The Solution
Kyle built an AI-invokable SQL Server profiler that lives inside his IDE. Instead of switching contexts to investigate database issues, he can simply ask his AI agent to check the current state of things.
Natural language input: "DTUs comin in hot! Check last 3 hours. Show top 3 offenders."
The skill maps that input sentiment to SQL parameters, queries the server's diagnostic views, and returns a clean synopsis of the worst-performing queries — complete with invocation counts, CPU waits, and tuning recommendations.
How It Works
The system combines PowerShell orchestration with targeted SQL scripts. When invoked, it:
- Parses natural language requests into query parameters
- Executes SQL Server diagnostic queries against system views
- Aggregates results into human-readable summaries
- Suggests specific tuning actions based on the data patterns
Tech Stack
Links
- Video Demo: YouTube →
- GitHub: seabass223/ai-skills →
- Original Post: Kyle on Substack →