BLADE Team Architect is a bilingual (English/French) team and build planner for
Xenoblade Chronicles X: Definitive Edition. It produces a sourced,
progression-safe general-purpose recommendation from the shared squad level
(the Avatar level), the Avatar's full class-rank map, each companion's
personal class rank, manual/AUTO roster choices, and Skell progression.
The app is a fully static Next.js site: all game data and the recommendation
engine run client-side. There is no backend, no account, and nothing leaves
your browser.
Highlights
- BLADE tactical dossier — a bilingual workbench with a compact, full-width
control deck above the results. Dark panels and thin cyan frames draw on XCX
menus; a faint upper hex pattern echoes Nintendo's manual framing.
Recommendation and member tabs scroll locally on small screens. Art
descriptions, researched build notes and Skell details stay closed until
opened, while the compact Skell summary remains visible. - Local configuration save — build settings and the configuration panel's
expanded/collapsed state are restored from browser storage after reload. - Availability before scoring — canonical class and companion records carry
exact Art/Skill unlock ranks, weapons and Skill-slot caps. Locked abilities
never enter the scorer. - Full Avatar progression — training a child records its ancestors at rank
10. Every trained branch contributes learned cross-class Skills; mastered
tier-4 classes add their weapon types and compatible regular Arts. - Companion-specific kits — all 22 Definitive Edition companions, including
Mia, use their own weapons, Arts, Skills, ranks and slot limits rather than a
generic Cross-class profile. Each runs one fixed, researched build
(src/data/companionBuilds.ts): at most two TP-consuming Arts — Cross keeps
its own three-Art budget — deterministic personal-rank substitutes, and
character-specific, sourced weapon/armor advice. A companion's loadout never
changes with Cross's class, weapons, teammates, team variant or combat mode. - Equipped-only companion scoring — team scoring reads each companion's
equipped Arts and Skills, and Skell aptitude counts equipped pilot Skills
only; Skell mode never silently rewrites a companion's ground kit. - Deterministic AUTO composition — manual picks remain in their exact slots.
AUTO slots enumerate level-eligible, non-duplicate companions and optimize
tank/support/damage specialist coverage before a tie-break informed by the
explicitly cited Escapist companion guide. - Shared squad level — one level drives the Avatar and every companion:
AUTO eligibility, loadouts and Skell pilot checks all use the Avatar level. - Definitive Edition progression — levels 1–99 are supported.
- Skell availability — frame recommendations require the Skell License and
the pilot's own level. Exact level-20, 30, 50 and explicitly enabled
postgame level-60 frames come with attribute-matched, availability-first
weapons and augments extracted from the wiki's semantic weapon database. - Source links in output — every selected Art and Skill links to the
canonical record used for its unlock check.
Scope
The result is a general-purpose heuristic, not a universal best-in-slot
claim. It verifies regular Arts and Skills represented in the canonical class
and companion tables. Encounter resistances, gear traits, owned inventory,
Affinity/signature Arts, mission restrictions, BP investment and execution can
change the best build. Skell weapons are the highest shop models matching the
strategy attribute at the frame's level band — an availability baseline, not
a best-in-slot claim; rare drop variants, superweapons and Ares weapons are
out of scope.
The tactical preference order lives in src/data/builds.ts. Factual unlock
tables live separately in src/data/progression.ts and
src/data/companions.ts; preference data cannot grant availability.
Companion builds are fixed and curated, not contextual. One canonical
Arts/Skills loadout per non-Cross character is kept in
src/data/companionBuilds.ts; unavailable abilities fall back to explicit,
ordered personal-rank substitutes, never a dynamic strategy re-ranking. Every
companion kit carries at most two TP-consuming Arts, strictly, including
substitutes — Cross's existing three-Art TP policy is unchanged. Unfilled
slots are legitimate: Irina deliberately equips three of her four Skill
slots because her remaining personal Skills depend on Sleep or Control that
the curated support bar never provides, and an inert fourth Skill is worse
than an empty slot. Character-specific
weapon and armor advice names verified manufacturer offerings, weight-category
trade-offs and availability gates; it is matchup guidance, not a universal
best-in-slot claim.
These curated builds are sourced recommendations, not a universal-optimum or
community-consensus claim. Research preferred Definitive Edition evidence and
validated original sources (Xeno Series Wiki game data, Game8, Nintendo Life,
Reddit and GameFAQs discussions) against each character's actual weapon pair,
unlock ranks and TP data, counting only independent recurrence — lists that
share one authorship count as one cluster, not consensus. Verified guide
errors were rejected rather than absorbed: weapon-incompatible
recommendations, missing TP markers and invalid Art names were dropped in
favor of game data. Where evidence stays sparse — most non-flagship
companions have no independently recurring full build — the final bar is an
explicitly editorial choice. Per-character evidence, decisions and gaps are
recorded in four research matrices:
core,
veterans,
specialists,
remaining.
Run locally
Requirements: Node.js 20.9 or newer.
npm install
npm run dev
The dev server prints its local URL (default http://localhost:3000).
Production build
npm run build # emits the static site to out/
npx serve out # optional: preview the production output locally
Deploy (free tiers)
Vercel
Zero configuration: import the repository on vercel.com.
Vercel detects Next.js, runs next build and serves the static export. No
serverless function is invoked at runtime.
Render
The repo ships a render.yaml blueprint: New → Blueprint on
render.com, pick the repo, done (free static site, no
cold start). Manual equivalent: New → Static Site with Build Command
npm ci && npm run build and Publish Directory out.
Any other static host (Netlify, Cloudflare Pages, GitHub Pages) works the same
way: build with npm run build, publish out/.
Project structure
app/ Next.js shell (layout, metadata, page, global styles)
src/data/ Static game data: classes, builds, Skells, AMs, i18n copy
src/lib/ Types and the recommendation engine (pure functions)
src/components/ React components; TeamBuilder is the client boundary
License
BLADE Team Architect is available under the MIT License.
Unofficial fan-made companion. Xenoblade Chronicles X and related names belong
to Nintendo and Monolith Soft. This project is not affiliated with, endorsed
by, or sponsored by them.
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 |
|