datasette-agent 0.2a0
Simon Willison · Simon Willison · 2026-06-10
Simon Willison releases datasette-agent 0.2a0, adding mid-execution user questioning that suspends agent turns as persistent forms and a human-approval-gated SQL query saving tool to the AI-powered Datasette database plugin.
Appears in
Extraction
Topics: ai-agentsopen-source-toolshuman-in-the-loopllm-tooling
Claims
- Tools in datasette-agent can now pause mid-execution to ask users yes/no, multiple-choice, or free-text questions via a ToolContext object.
- Suspended agent conversations persist to the internal database and survive server restarts, with stored answers replayed on re-execution.
- A new save_query tool allows the agent to save generated SQL as a Datasette stored query, but always requires explicit human approval before anything is written.
- The ask_user() feature was enabled by a new LLM alpha Simon Willison built with the help of Claude Fable 5.
Key quotes
Tools that declare a context parameter receive a ToolContext object, and await context.ask_user(...) can ask a yes/no, multiple-choice (options=[...]) or free-text (free_text=True) question.
While a question is unanswered the agent turn suspends: the question renders as a form in the chat UI and persists to the internal database, so suspended conversations survive a server restart.
Saving always requires human approval - the agent shows the full SQL plus the proposed name, database and visibility, and nothing is stored until you click Yes.