Demo Mod — Export / Import / Promote

Scope: Session dials and piece overrides leave the HTML mock (and later W_TFA_DemoMod) as one file per concern, then promote into cooked data (DT / UDA / CSV).
Interactive: tfa_gdd_html/figures/demo_mod_mock.html
Schema: tfa-demo-mod/v1
Promote CLI: python promote_demo_mod.py <exports> [--out data/promoted/<run>]

Why

Lab formulas, Ult/Energy rates, gunfeel, and doll overrides must not stay trapped in the Mod session. Export → review → Promote to data (team-only). Shipping builds read cooked rows only (FS-GUN-012 · combat-economy design §6).

File kinds (own files)

Kind	Filename pattern	Payload
`lab-formulas`	`tfa-lab-formulas.json`	Attribute per-pt + encumber dials (FS-PRG-002)
`lab-economy`	`tfa-lab-economy.json`	Ult / Skill Energy streak dials (FS-CMB-007)
`lab-flow`	`tfa-lab-flow.json`	Asymmetric Flow · second meter · Burst Storm / Singularity (FS-CMB-004/005)
`lab-parkour`	`tfa-lab-parkour.json`	`WallRunMaxDuration` · `MaxWallDashDistance` · `MaxWallTurnDegrees` · lookahead distance/samples · DEX %/pt on duration/dash (FS-MOV-004)
`lab-raw`	`tfa-lab-raw.json`	STR…WIS integers (lab sandbox)
`lab-encounter`	`tfa-lab-encounter.json`	alvl / dlvl / ARL / Difficulty / PowerΔ / CtH dials
`piece`	`tfa-piece-<seat>-<id>.json`	One committed (or draft) doll override — plate · affixes · gunfeel · kit/quality/sockets
`manifest`	`tfa-export-manifest.json`	Index of split files + fillOverrides / ammo meta

Envelope (every file):

{
  "schema": "tfa-demo-mod/v1",
  "kind": "lab-formulas",
  "exportedAt": "ISO-8601",
  "scenarioId": "…",
  "payload": { }
}

Everything (split files) downloads the manifest, then the six lab files (economy · flow · parkour · raw · formulas · encounter), then each committed piece — browsers may ask to allow multiple downloads.

Mock UI

OV36 list: Export dropdown + Download · Import file… (multi-select JSON).
OV37 edit: Export this (current draft, including unsaved gunfeel / formulas).

Import merges into session state (lab layers or committed[key]). Does not auto-cook Unreal data.

Promote path (CLI — Done)

# validate only
python promote_demo_mod.py data/demo_mod_exports/fixtures --dry-run

# write cooked targets
python promote_demo_mod.py data/demo_mod_exports/fixtures --out data/promoted/smoke
python promote_demo_mod.py path/to/exports   # → data/promoted/<utc-stamp>/

Export kind	Cooked artifact	Ship?
`lab-economy`	`DT_TFA_CombatEconomy.json`	Yes — Ult/Energy rates (FS-CMB-007)
`lab-flow`	`DT_TFA_FlowSingularity.json` (or DA Burst Storm table)	Yes — Asymmetric Flow / second meter / Burst Storm (FS-CMB-004/005)
`lab-parkour`	`DT_TFA_ParkourLabDefaults.json` → parkour DA	Yes — WallRun 3.0 s / dash 6.0 m / turn 45° / lookahead 2.0 m·4 + DEX on duration/dash (FS-MOV-004)
`lab-formulas`	`DA_TFA_AttrFormulas.json`	Yes — per-pt feeds (FS-PRG-002)
`lab-raw`	`DT_TFA_LabAttrRaw.json`	No — character sandbox only
`lab-encounter`	`DT_TFA_EncounterLabDefaults.json`	No as global — area owns Difficulty/ARL
`piece`	`kits/<seat>__<id>.json` + `gunfeel/DemoKitGunfeel.csv`	Yes — kit fill / gunfeel columns

Each run also writes REPORT.md + promote-manifest.json. Fixtures: data/demo_mod_exports/fixtures/.

UE handoff (still Need)

Step	Notes
Same schema from UMG	`W_TFA_DemoMod` Export writes the same `tfa-demo-mod/v1` files
Editor import	Team imports promoted JSON/CSV into DT/UDA / `UDA_TFA_GunFrames` columns
External	Mod gated OFF — no export RPCs; runtime uses cooked data only

Non-goals

Silent C++ defaults from a half-imported file · client-owned damage/cadence · replacing area-authored Difficulty with a global JSON drop · auto-cooking .uasset without the editor.