
Datasette Agent 0.3a0 introduces user approval prompts for write operations via the execute_write_sql tool. The chat terminal now supports approvals and includes --unsafe mode for auto-approving actions. Users can interact with databases via commands like datasette agent chat content.db -m gpt-5.5 --unsafe. A $10/month sponsorship offers a monthly LLM development digest.
Tap to vote and see what everyone thinks.
Summary by ByteBrief
Rubrik Now Available as AI Agent, Company Announcement