nikke-sim

Solo raid (180 s) damage simulator for NIKKE — a CLI and a website running the same engine. Input 5 Nikkes in slot order, get expected damage per unit from a full burst-rotation simulation. See DESIGN.md for architecture, nikke-damage-formula.md for the formula source, future.md for the roadmap.

Setup

npm install
# .env needs DATABASE_PUBLIC_URL (Bakery Bot's Postgres)
npm run sync          # DB + nikke-synergy API → data/*.json (rerun after roster updates)
npm run sync:skills   # blablalink per-skill-level values (enables skill levels < 10)

Website

npm run web           # dev server
npm run web:build && npm run web:preview   # production build → http://localhost:4173

Static site, no backend — the sim runs client-side (deploy dist/ anywhere). Features: 5 slot pickers with portraits, boss element + boss DEF, core visibility presets (No Core → 100% / custom), per-character OL0/OL5 gear + Doll 15 + cube (Resilience / Bastion / Other at L7/L10/L15/custom), damage + % share per unit, rotation log, modeling notes. All skills factored at 10/10/10.

CLI

npm run sim -- liter crown naga modernia alice --element Iron --doll yes --ol 5 \
  --cubes "resilience,resilience,bastion,bastion,resilience" --boss-def 8000 --core-rate 0.5
npm run sim -- --list [filter]      # browse slugs
npm run sim -- --coverage           # parser health across the roster

Options: --element, --boss-def, --doll yes,no,… (per slot; 1 entry = all), --ol 0,5,… (per slot), --level (default 400), --copies 0-10 (default 3), --core-rate 0-1, --no-range, --duration, --rotation.

Loadout (per-slot lists in slot order; one entry applies to all 5):

• --cubes resilience@15,bastion@7,-,… — harmony cube per slot (real values from cubes.md: flat ATK + elemental damage from L5 + effect)
• --lines "elem*4+atk*4;;;ammo*2;" — OL lines, type[*count][@value] (max rolls from data/ol-lines.json)
• --skill-levels "10/4/10,…" — per-skill levels using exact blablalink per-level tables
• --best-ol <1-5> — greedy best-OL search for that slot: marginal % damage per max-roll line, cap 4/type

How it works

• 60 fps frame simulation: weapon fire cycles (AR 12/s, SMG 20/s, SG 1.5/s, MG 60 rounds/s with 3.75 s spin-up, RL/SR charge), ammo/reload/Bastion refunds, burst gauge (rl3/3 %/s), stage I→II→III→full-burst rotation gated by real cooldowns (CDR modeled, incl. once-per-battle Λ refunds).
• Stats reconstructed exactly from the DB (base_stats + synchro curve) + gear ATK by class/OL + doll + cube flat ATK.
• Skills: prose parser → structured effects in formula buckets, hand-verified overrides in src/skills/overrides/*.json take precedence (all Prydwen bossing-tier S+ units are hand-verified; see docs/override-guide.md for authoring, scripts/kit.ts + scripts/validate-overrides.ts for tooling). Unmodeled effects are surfaced as modeling notes — never silently dropped.
• Λ units (Red Hood) burst at any stage by default; pin them with --lambda-as (CLI) or the "as B1/B2/B3" pills (site).

v1 assumptions

No enemy debuffs on allies (full HP assumed), expected-value crits, always in effective range. Boss DEF and core-hit rate are inputs. Ally-applied Damage Taken debuffs ARE modeled. Parts/pierce are v3 (future.md).