datasette-agent 0.2a0

2026-06-10 · Source: Simon Willison's Weblog · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Data Science & Analytics, Software Development & Engineering · Depth: Intermediate, quick

Summary

datasette-agent 0.2a0, released on 10th June 2026, introduces significant enhancements for its LLM-powered agent capabilities within Datasette. Key features include the ability for tools to ask users questions mid-execution via an `ask_user()` function, supporting yes/no, multiple-choice, or free-text inputs. This interactive questioning suspends the agent's turn, rendering a form in the chat UI, and persists across server restarts. Additionally, a new built-in `save_query` tool allows the agent to save generated SQL as a Datasette stored query. This saving process always requires explicit human approval, displaying the full SQL, proposed name, database, and visibility before storage. The `ask_user()` feature was enabled by a new LLM alpha developed with Claude Fable 5.

Key takeaway

For data scientists or AI engineers building LLM-powered data tools, `datasette-agent 0.2a0` offers crucial advancements in agent-user interaction and controlled data persistence. You should consider integrating this version to develop more robust, user-guided data analysis agents that require explicit human approval for critical actions like saving queries, thereby increasing reliability and trust in automated processes.

Key insights

LLM agents can now interactively seek user input and save approved queries, enhancing control and persistence.

Principles

• Tools should ask for user input before performing side effects.
• Human approval is critical for agent-initiated data modifications.

Method

Tools declare a `context` parameter to receive a `ToolContext` object, enabling `await context.ask_user(...)` for interactive questioning and re-execution.

In practice

• Implement interactive agent workflows requiring user clarification.
• Enable agent-generated SQL to be saved as Datasette stored queries.

Topics

• datasette-agent
• LLM Agents
• Datasette
• User Interaction
• SQL Queries
• Claude Fable 5

Code references

• datasette/datasette-agent

Best for: AI Architect, AI Product Manager, AI Engineer, Data Scientist, Software Engineer

Related on AIssential

• Datasette Agent — Datasette Agent integrates LLMs with Datasette for conversational data querying and extensible AI assistance.
• datasette-agent-sprites 0.1a0 — datasette-agent-sprites 0.1a0 integrates Datasette Agent with Fly Sprites for secure, sandboxed command execution.
• humanlayer / 12-factor-agents — Reliable LLM agents require robust software engineering principles, not just advanced AI frameworks.
• The Newsroom Protocol: Why Autonomous AI Agents Are a Ticking Time Bomb in B2B Databases — Autonomous AI agents with direct database write-access are a critical security and operational risk due to their probabilistic nature and susceptibility to prompt injection.
• Text to Data Products: Kaarvi’s End-to-End AI for Ingestion, Quality, and Dashboards — Kaarvi AI employs a multi-agent, parallel LLM architecture to automate data readiness, significantly improving reliability and efficiency in data management.
• How OpenAI Built Its Data Agent — The reliability of AI agents at scale stems from strong data foundations and carefully engineered context, not complex agent architectures.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Simon Willison's Weblog.