Datasette

Part of MOOLLM · skills/datasette/

Datasette (datasette.io) is the CLI and runtime for publishing SQLite files as a read-only web application and JSON API. It sits on top of the same .db file format as the sqlite engine: this skill is the tooling and ecosystem layer (serve, metadata, plugins, CORS), not a second database engine.

sqlite → datasette: you pass the same path as sqlite3 my.db. Datasette does not build a second on-disk database for serving—read-only HTTP over that file. That is the usual sense of “zero copy” live publishing here: no duplicate .db artifact; refreshes see committed data from other writers when the file uses a normal WAL setup. For a static snapshot, serve a copy or use --immutable on a file you do not change.

What you get

• datasette — local web server; table browse, filters, Custom SQL, export links.
• Metadata YAML — per-database titles, facets, sortable columns, canned queries (named SQL with optional :parameters).
• --crossdb — attach multiple SQLite files; run SQL that references more than one database (see Datasette docs for exact SQL shape).
• Plugins — ecosystem extends CSV export, maps, full-text helpers, etc. (Plugin directory).
• JSON API — append .json to many URLs for programmatic use (JSON API).

Official documentation: https://docs.datasette.io/

Install