NOTES FROM THE LAB
Curiosity, with receipts.
What we can observe, what the community reports, and what still needs a test. This is a working parameter catalog, not an official JJS specification.
Code validity is one checkpoint.
The editor checks the compressed format and JSON structure. Those checks cannot establish whether JJS will accept a code or how a changed move will behave.
Start with an unchanged export in your private server. Then change one parameter at a time and record the result.
Read the confidence label.
Observed in sample A value exists in the supplied export; its meaning may still be unknown.
Community reported A third-party definition or player account, not a game guarantee.
Manually tested Requires an actual private-server result. No gameplay claims below carry this label yet.
PARAMETER CATALOG
Start with what is there.
| Parameter / context | Evidence and editing guidance | Confidence / source |
|---|---|---|
ATTACK TYPEHITBOX nodes | The community library lists An imported value outside this list is retained. Test each candidate independently. | Community reportedFormunaGit / libjjs types ↗ Observed 2026-08-28 |
NeighborDAttack type value | The supplied Chase moveset contains this value in its SpotDefault hitbox. Its presence does not prove wall penetration, a special targeting rule, or compatibility with every node. | Observed in sampleSupplied Chase export Observed 2026-08-28 |
DAMAGESTUNHITBOX nodes | The sample includes numeric damage and stun values, including 2.5 and 1 in SpotGo. Keep the JSON type. Units, limits, and interactions with other nodes remain game-test questions. | Observed in sampleSupplied Chase export Observed 2026-08-28 |
BLOCKABLESINGLE TARGETHIT USERHIT RAGDOLLIGNORE WAKEUP | These flags appear as booleans in the sample hitboxes. Guided controls retain that type. Their names are clues, not verified descriptions of all runtime behavior. | Observed in sampleSupplied Chase export Observed 2026-08-28 |
SIZEPOSITIONROTATIONHITBOX nodes | The sample represents these vectors as strings such as "8, 8, 8". Do not replace them with arrays or numbers just because they look numeric. Other node kinds can use different types for similarly named fields. | Observed in sampleSupplied Chase export Observed 2026-08-28 |
COOLDOWNCharacter slot / SETCD node | The sample has a slot cooldown of 24 and a separate SETCD node with 8. They are separate fields. The editor never assumes changing one should change the other. | Observed in sampleSupplied Chase export Observed 2026-08-28 |
STATE: "IFrame"Existing STATE node | The sample contains an IFrame state with VALUE: 1 and TIME: 3. Advanced mode can edit those existing scalars. No permanent invulnerability preset or guarantee follows from their presence. | Observed in sampleSupplied Chase export Observed 2026-08-28 |
DATACharacter-slot envelope | The supplied outer document is an array of character slots. Each slot’s DATA contains JSON encoded as a string. The editor preserves every slot and keeps DATA a string on export; unfamiliar envelopes remain available for inspection. | Observed in sampleSupplied Chase export Observed 2026-08-28 |
CAN COLLIDEVISUAL nodes | This flag occurs on visual-effect nodes in the sample. That is not evidence of a character or projectile noclip switch. A separate player thread reports projectiles stopping at walls, with speculative suggestions in its replies; it does not establish a reliable fix. | Observed in sampleCommunity reportedChase export · Player wall report ↗ Observed 2026-08-28 |
Keep the experiment small.
- Verify the control.Import the unchanged baseline. Note the test date, private server, move, and exact setup. If the baseline fails, stop and investigate the format.
- Change one field.Generate a short list of candidates in Advanced mode. Keep the same target and setup. Save the exact code and diff for each attempt.
- Record two answers.Did the game import it? What actually happened during play? Record those separately. One successful import is not evidence for an invulnerability or wall-clipping claim.
The supplied sample is used as a regression fixture. Community sources are attributed research references; public availability is not permission to redistribute someone else’s moveset. Game updates may change behavior.