Simon Willison's datasette-agent now lets AI assistants modify your databases—but with guardrails. Version 0.3 adds an approval step that displays exactly what SQL will execute and what permissions are required before the AI writes anything. You can also run it in unsafe mode to skip confirmations, for when you trust the model completely. It's a practical approach to giving LLMs database access without handing them the keys to your data.

Listen to this story

Hear this and more stories in a personalized audio briefing.

Open The Chonkerton