The database engine, 2,574 files of C. Grouped along its source tree: the backend (910 files), the interfaces, bin, and tests.
The largest areas IngestMD found
On a flat C tree the names follow the directories. A local model can rename them, see the raw-to-named demo on this page.
How it was made
IngestMD scanned PostgreSQL on the laptop and grouped its files into coherent areas in ~4s, with no model and no network. The result is deterministic, so the same repository produces the same map every time, and no two areas share a name. From here you load just the area you need as focused context for an AI assistant, instead of the whole tree.
Map your own repository
Point IngestMD at any folder and get the same kind of map, on your machine, in seconds.
Download IngestMD