A combination of --jsonargs and --slurpfile. Slurps all json files passed as a command line argument, and makes them available under $ARGS.named[filename].

$ jq -n '{"a":"a"}' >a.json
$ jq -n '{"b":"b"}' >b.json
$ jq -n --experimental-fileargs '$ARGS.named' *.json

{
	"a.json": [
		{
			"a": "a"
		}
	],
	"b.json": [
		{
			"b": "b"
		}
	]
}

Using this over an input builtin allows easy access to filesystem functionality such as find and globbing, without having to implement them as jq builtins. Files can also be written-in-place when using snapshots ( #3165 ), since all inputs are read before any writes happen.

Depends on #3165 for tests, marking as draft until that one is merged.