Query a SQLite database

No. The file is read into memory and opened by sql.js, a build of SQLite compiled to WebAssembly that runs inside the page. The query executes on your device, the results are formatted there, and the database is closed and freed from the WebAssembly heap when the run finishes. Nothing is sent to a server and there are no accounts.

You can run the statement, but the change goes nowhere. The database is opened from a copy in memory and no file is written back, so a write statement runs against that copy and is discarded when the run ends. You get "statement ran, no rows returned" because such a statement produces no columns. Treat this as a read tool.

Because that is the cap, and the tool tells you: "first 1000 rows only" with a suggestion to add LIMIT or WHERE. Rows past the cap are never even read, since the statement is stepped row by row rather than materialised in full, which is what stops a million row table from building a million line string in your browser.

No, one statement per run. The default is a query that lists the tables in the file, which is a good first thing to run when you do not know the schema. Run your statements one at a time, reading the output of each before the next.

Because the bytes you gave it are not SQLite. That message comes from SQLite itself, prefixed with your file name, and it appears when the statement is prepared rather than when the file is opened, because SQLite only reads the header at that point. A zero byte file is caught earlier and refused with "file is empty", since SQLite would otherwise open it as a valid empty database and answer your query with silence.

A NULL prints as NULL and a BLOB prints as its size, in the form <blob 2048 bytes>, because printing the raw bytes would produce a wall of numbers. Everything else prints as text, in a table whose columns line themselves up; nothing is padded by hand. A long text value stays on one line and the table scrolls sideways rather than wrapping it.