The content is stored into the data*.tar in the crx-quickstart/repository/workspaces/crx.default directory IIRC.
If you know how to you could potentially access the content within the tarball. Just remember that by default content bigger than 4k will end-up into the datastore.
While I agree, that you can potentially access the data directly on the filesystem, I strongly advise not to do this. It is an implementation detail of CRX and change without further notice. It's the same reasons, why you don't access an oracle database file directly on the filesystem, but only via its documented API.