User Flow Map — gskillpacks.com Deck Creation

/user-flow-map was invoked to map the full deck-creation flow on gskillpacks.com — landing on the table through deck completion and CLI unlock, covering preset (canonical) decks, customization (swap/overlay), and custom decks — and to pressure-test whether phase-labeled slots mirroring the canonical card list make building feel like a chore versus user-defined phases. That second question was explicitly routed here by Amendment 2 of the animation design plan.

Evidence behind load-bearing decisions

Confirmed in the terminal interview on 2026-06-10 ("Confirmed as-is"). The slot-model direction was answered verbatim: "we have preset decks which users can modify as needed with default phases." Mobile branch notes were explicitly included.

The question

Do phase-labeled slots that mirror the canonical card list 1:1 make building a preset deck feel like a chore with no choices — and should phases instead be user-defined?

Analysis

The chore risk is real but concentrated, not structural. Slot-by-slot interaction cost per preset deck: VARD 3 taps, ORD 3, Devtool AFPS 8, Game AFPS 11, Business AFPS ~35. Transcription fatigue lives almost entirely in Business AFPS; the locked UX already mitigates it there with stacked pack-slots and per-fan add-all.

The genre distinction that matters: netdecking vs brewing. Deck-builder games separate copying a known list (netdecking — players want one-click import) from composing your own (brewing — players want freedom). Preset decks are netdecks: nobody wants to place a known-good list card by card. The chore is not the phase structure — it is being forced through per-card placement when the user has already made their one real decision ("I want this deck"). The fix is speed, not loosened structure.

Why user-defined phases on preset decks would cost more than they give. (a) The one-line npx skillpacks install-deck <slug> contract depends on the preset staying canonical; editable phase structure forces every deck onto the explicit-pack-list path. (b) "The deck IS the pipeline" is the product's teaching mechanism — phase columns read left-to-right as execution order; user-defined phases delete the lesson. (c) Graduation arrows (VARD→Business AFPS, ORD→Devtool AFPS) connect canonical structures; they have no meaning between freeform trays.

Resolution (encodes the user's direction)

Consequence for the locked animation contracts: the card-flight target contract (slot identity/position) is unchanged — phase slots remain the targets. Only the new deck-level Collect-all needs animation-plan review; movement mechanics (optimistic commit, clone overlay, snap-to-end) are untouched.

Entry points & preconditions

Happy path (preset deck, Evaluator)

Interactive walkthrough (flow states, not motion)

Flow diagram (all screens & transitions)

Alternate paths

Decision points

Kept outside this flow's scope: /follow and /admin/newsletter (unchanged), legacy showcase pages (archived once the deck-builder ships, per the idea brief). /catalog and /packs become redirects (E6).

Per-screen action/state matrix

Failure & recovery

Handoffs & external/manual dependencies

Structural regions only — no styling, spacing, or visual design. Those belong to /ui-interview and later stages.

S1 Table (/)

┌─ brand strip ──────────────────────────────────────────────┐
│                                                            │
│  BLUEPRINT ROW (primary)                                   │
│  [VARD 2/3] [ORD] [Biz AFPS] [Devtool AFPS] [Game AFPS]    │
│      └────traction──→ [Biz AFPS]   (faint arrows)          │
│  [ + Custom deck ]   ← first-class sixth tile (§3)         │
│                                                            │
│  PACK SHELF (secondary, below the fold acceptable)         │
│  ( 7 sealed boosters, horizontal scroll )                  │
└────────────────────────────────────────────────────────────┘

• Blueprints carry progress badges from localStorage; the badge is the resume affordance (B7).
• Shelf packs open inspect-only from here (B5); no deck context implied.

S2 Preset builder (/deck/[slug])

┌─ deck name (editable only as custom variant)      [✕] ─────┐
│  PHASE COLUMNS, left→right = execution order               │
│  ┌ Scan ─┐  ┌ Align ─┐  ┌ Ship ─┐                          │
│  │ ghost │  │ FILLED │  │ ghost │   ← ghost = suggestion   │
│  └───────┘  └────────┘  └───────┘     outline + /command   │
│  Business AFPS: column = stacked pack-slot w/ counter      │
│  ─ overlay row (freeform, any card) ───────────────────    │
│  [ Collect all N ]                ← deck-level (B1)        │
│  CLI PANEL (persistent): 🔒 2 more to unlock │ [copy]      │
│  ─ condensed shelf strip (packs reachable in-context) ──   │
└────────────────────────────────────────────────────────────┘

• CLI panel visible from first paint in locked state — the destination is never a surprise.
• Tapping a ghost glows the source pack in the condensed shelf strip; the fan opens over the builder with slots still visible (flight targets).

S3 Custom builder (/deck/custom)

┌─ [name your deck…]                                [✕] ─────┐
│  ┌ Phase 1 (rename) ─┐  [+ add phase]                      │
│  │ empty freeform    │   phases: add / rename / remove     │
│  │ slots (any card)  │   (remove asks if cards inside)     │
│  └───────────────────┘                                     │
│  ─ overlay row ─────────────────────────────────────────   │
│  CLI PANEL: 🔒 add a card to unlock │ explicit pack list    │
└────────────────────────────────────────────────────────────┘

S4 Pack fan drawer

╔═ bottom sheet (drag-dismiss) ══════════════════════════════╗
║  pack name · fan of cards (flip = inspect, tap = collect)  ║
║  [card] [card] [card] [card]      [ Collect all 4 ]        ║
║  inspect-only mode: tap → card detail, "add to deck" → S7  ║
╚════════════════════════════════════════════════════════════╝

S5 Card detail / S6 Output / S7 Deck picker

S5: [ card art/front | stats back: type, benchmark grade,    S6 (deck back):
      version, /command, pack ]                               npx skillpacks install-deck vard   [copy]
    [add to deck]  (standalone: [Open on the table])          + install code-review (overlay)
                                                              [⬇ project.json] [share] [keep editing]
S7: "Add /benchmark-card to…"  ▸ VARD (2/3)  ▸ ORD  ▸ + new custom deck   [cancel]

Mobile branch notes (wireframe level)

• Slot strip pinned above the BottomSheet: when the fan is open on mobile, the active deck's slots condense to a pinned horizontal strip so flight targets stay visible (carries the locked animation constraint into layout structure).
• Tap-everything: all collection and navigation is tap-based; the tear gesture gets a tap-to-open equivalent (already in SealedPack). No interaction is drag-only.
• Phase columns become vertically stacked phase rows on narrow screens; left→right order becomes top→bottom; Business AFPS pack-stacks stay collapsed by default.
• Output panel becomes a full-width sheet; the command line wraps with a copy button adjacent, not trailing off-screen.

Explicit non-goals

• No visual styling, color, component design, or design tokens (downstream: /ui-interview, /ux-variations --layout-mode).
• No animation mechanics — motion contracts stay owned by the approved animation plan; this flow only fixed slot targets and added the Collect-all trigger to be amended there.
• No accounts, trading/progression systems, or server-side state.
• No CLI behavior changes; install-deck semantics are an input, not an output.
• No production architecture, database schemas, or implementation plans.
• No mobile-specific spec beyond the wireframe branch notes in §10.

Destination approval and mutation scope are separate decisions below (the destination is a path choice; the file-change set includes the index update and page confirmation).

The AFPS design route after an approved flow map is per-screen content requirements, then layout exploration: /ui-interview --requirements-only deck-creation → /ux-variations --layout-mode. The ui-interview skill is in the product-design pack, which is not installed (only user-flow-map and brainstorm are individually enabled) — so the route has an install prerequisite: /pack install ui-interview (or the full pack). A second, smaller follow-up is the animation-plan amendment for the Collect-all batching trigger (§11).