SQLite browser

The repository's SQLite files in a tile, read-only. Choose a file: its tables and views are listed with their row counts; choose one to see its first 200 rows in a grid (numbers right-aligned, nulls marked). Write any query in the box and run it with the button or Ctrl+Enter.

It finds the .db, .sqlite and .sqlite3 files in the repository (outside node_modules, .git and target), git-ignored ones too, as regular files only: find never follows a symbolic link. It opens them with sqlite3 -readonly -safe -nofollow -json: -readonly refuses every write, -safe refuses sqlite3's commands that run programs or attach other files, and -nofollow refuses a path through a symbolic link. Right before each query it checks again that no folder on the way to the file, nor the file, is a link, so a file or folder replaced by a link after it was listed is refused. A query starting with . or - is not sent at all. An agent names a database only from the files the tile lists, never a path outside the repository.

hive plugins install @dip497/sqlite-browser
hive plugins enable @dip497/sqlite-browser
hive ctl open-tool @dip497/sqlite-browser/browser

An agent asks it, while its tile is open:

hive ctl plugin call @dip497/sqlite-browser tables
hive ctl plugin call @dip497/sqlite-browser query '{"sql":"SELECT status, COUNT(*) FROM orders GROUP BY status"}'
hive ctl plugin call @dip497/sqlite-browser query '{"sql":"SELECT * FROM users LIMIT 5","file":"data/app.db"}'

It needs sqlite3 3.37 or newer (for -safe). It asks for no permissions.

How it was made

Built by a coding agent (Claude) from the tool-plugin contract (the brief hive plugins new tool writes, and the tool plugin guide). It uses host.run of git ls-files, find and sqlite3, and useTileState to keep the chosen file, table and query. Building the queries and reading sqlite3's answers are in lib.js, tested against a real database.