How paldb sources its Palworld data

The pipeline

1. Palworld’s pak files are downloaded from the game’s own depot and unpacked with a mapping file matched to that exact build.
2. The DataTables are exported as JSON — 483 tables in the current build, of which 36 carry data this site publishes.
3. An ingest step reshapes them into the compact tables the site ships. It does no rounding, no filling and no inference: if the game does not state a field, the field is absent and the page renders the absence.
4. A validation step refuses the build if a table has lost a large fraction of its rows, which is what a broken extraction looks like from the outside.
5. The snapshot is archived in full before anything overwrites it, so every build can be compared against every other one — see the change explorer.

The freshness stamp on every page is the game build’s timestamp, not the time this site was rebuilt. That means it can go stale and show it, which is the only version of a freshness claim worth making.

What the labels mean

Game data

Exact values extracted from Palworld's game files.

How paldb sources its data

Game files

Read directly out of a DataTable. Exact, and the vast majority of this site.

PalDB calculation

Derived from extracted game data using the formula shown.

How paldb sources its data

Calculated from game files

Arithmetic over extracted values. The formula is printed wherever the number is.

Verified

Checked by hand against the game.

How paldb sources its data

Manually verified

Extracted from the game, but from somewhere other than a DataTable — a string table, say — and mapped by hand.

Community

Community-documented and widely agreed, but not extracted from the files.

How paldb sources its data

Community verified

Not in the files. Established by players, and labelled as such wherever it appears.

Not extracted

The game files do not state this value.

How paldb sources its data

Source unknown

The game does not state it and nobody has established it reliably. paldb leaves these blank rather than guessing.

Everything on paldb that is not first-party — all 5 of them

This list is short on purpose, and it is the part of this page worth reading. If it ever gets long, the claim at the top of this page has quietly stopped being true.

What the files genuinely do not contain

Some things are absent rather than hidden, and no amount of extraction will produce them. There is no egg hatch table anywhere in either the client or the server pak — we checked both, and they contain an identical 915 DataTables. There is no passive inheritance table. Merchants, NPCs and nests carry no extracted coordinates. Terrain flatness is not a number the game writes down.

One class of absence turned out to be recoverable. Ore nodes, gatherables, chests and fishing spots are not in any DataTable — they are actors placed in the world’s 9,977 level cells, which is why every Palworld map site hand-plots them. paldb now sweeps those cells directly with the same extraction tooling, so the resource layers on the map are exact positions from the game’s own files, refreshed with each build like everything else.

Where a competitor publishes one of these as an exact figure, it is a reverse-engineered estimate. That may well be a useful estimate — but it is not the same kind of thing as a value read out of the game, and paldb will not present the two as though they were.

Historical data

paldb retains a complete snapshot of every game build it has ingested — 2 so far — so that stat changes can be measured rather than remembered. Every entity page carries its own history, and the change explorer lists every field that has moved between builds, with the table it came from.