[KongchangAI]
Productdatasette-agent

datasette-agent

An agent plugin for Datasette that builds on the model configuration of datasette-llm, enabling an AI agent to drive conversational interactions with a database.

Timeline (last 90 days)

Nov 16

Putting the question box on the homepage means natural-language querying has shifted from an optional plugin to one of Datasette's default interaction entry points.

Unverified50%
Nov 16

The input box supports hash-mentions syntax, i.e. using # plus a table name or query name to precisely reference a database object, and has prefix-matching autocomplete.

Unverified50%
Nov 16

According to the release notes screenshot, typing #re shows the repos and releases tables in the autocomplete list.

Unverified50%
Nov 16

datasette-agent is a plugin that lets users ask questions of Datasette data in natural language.

Unverified60%
Nov 16

Version 0.6a0 added the agent question box to the Datasette homepage, below the search box.

Unverified50%
Nov 16

The 0.6a0 release notes position it as a cumulative set of small fixes.

Unverified50%
Nov 16

The Datasette Agent area added to the homepage in 0.6a0 contains a text input box and an ask button.

Unverified50%
Nov 15

In the official example, datasette-agent uses the lmstudio-qwen model to run queries against the content database, obtaining a repo_count of 200 and a release_count of 1889.

Unverified50%
Nov 4

0.5a0 is still an alpha version, and its features and stability may still be iterating.

Expired60%

All Facts (8)

Source Articles