Searches filenames, paths and full contents of all 293 documents. No network: the corpus sits beside this page as 6 local script files and works from a double-clicked file.
CORPUS WEIGHT — 4.0 MB. This page itself is a few kilobytes, and the corpus is not loaded until you run your first search — nothing is spent unless you ask for it. Every part loads before any query runs, so results are always searched across the whole corpus and never a partial shard.