Advisory 01 — the art plan is sequenced against the product's own wager

From: WootBuild · To: docade · Date: 2026-08-17 Status: ANSWERED 2026-08-17.

Written before producing anything, per 00-CHARTER.md § The studio advises. Every claim below cites docade's own files so it can be checked rather than trusted.

The finding

docade has zero pieces of product artwork today, and the product works.

public/ and assets/ contain eleven image files:

Not one is docade art. The app has been built to a working core loop on deliberate fallbacks — src/lib/prizes/art.ts draws gradient-and-emoji prize cards rather than 404ing, and its own comment calls that "the permanent path for most prizes", not a stopgap.

That is not a gap to close. It is evidence about what is load-bearing, and it does not support the current phasing.

The phasing inverts the product's wager

00-PRD.md §1 states the bet plainly: the category is not new, and docade wins on "depth of gamification — a real two-currency arcade economy, a prize wall, streak-gated blind-box pulls, and a collection to build, rather than gold stars on a fridge."

Against that, here is where the art budget sits:

103 of 204 pieces — half the catalog — ship before a single collectible exists. The collection is Phase 4. The crane is Phase 3. The differentiator lands last, and Phase 1's art is the chrome around the chore half — which is precisely the "gold stars on a fridge" the PRD says docade is not betting on.

A household enrolling at the Phase 1 exit criterion sees a beautifully iconed chore list and none of the arcade. That is the version of docade most likely to be judged, and it is the one carrying no evidence of the wager.

Three specific recommendations

1. Move the collectible probe ahead of the glyph batch

Already half-agreed: reference sheet piece 2 (octopus ×2 colorways) settles Q44, and Q45 is explicitly blocked behind it. Extend that logic. The plush lane is simultaneously the highest-risk (its economics are unproven), the highest-value (it is the wager), and the longest-lead work. The glyph lane is the lowest-risk: 58 monochrome 24px pieces that any competent icon set could stand in for while the collectibles are proven.

Risk and value both say plush first. Only the phase numbering says glyphs.

2. Cut the 15 prize illustrations to ~5, on docade's own reasoning

src/lib/prizes/art.ts:

"A parent-created prize will ALWAYS outnumber the illustrated ones — the whole point is that a household invents its own — so a kind-based fallback is not a stopgap for the missing art, it is the permanent path for most prizes."

If that is true — and it is the architecture that shipped — then 15 bespoke prize illustrations serve the screenshot, not the product. The four generic kind-fallbacks carry the real wall. Recommend the four kind fallbacks plus a small number of hero prizes for the store listing, and let the gradient system do what it was designed to do. Saves ~10 pieces of Phase 1 art with no product loss.

3. The 58 glyphs are one system, not 58 illustrations

At 24px with no colour, the requirement is family coherence — laundry must be distinguishable from fold-clothes because they sit in the same list (reference sheet, piece 7). Sibling coherence cannot be achieved by generating siblings independently; it has to be drawn as a set, with a shared stroke weight, corner radius, optical size, and metaphor vocabulary.

This changes the production method, not just the order: glyph should be an authored vector system, and WootBuild should propose a grid and stroke spec before drawing any of the 32. It is also worth asking whether 32 task icons are needed at launch when a star key already exists as a generic.

What we are not challenging

Recorded so the endorsement is as explicit as the criticism:

• The reference-sheet-first method is right. "Style drift discovered at asset #40 costs the whole batch" is correct, and thirteen pieces chosen to each govern a batch is a good sheet, not an over-thorough one.
• Filenames as foreign keys is right, and WootBuild will enforce it: art deliver refuses any path that is not a declared manifest key.
• "Derive silhouettes, don't draw them" is right — 18 images for zero art budget, and they can never drift from the source. WootBuild will do this in sharp.
• The placement harness is a genuine prerequisite, not a nice-to-have, and the gap is ours rather than docade's. Being built into the WootBuild console now.
• Refusing to invent a key for piece 12 was right. A request that cannot become a brief should escalate, not get guessed.

Net effect if all three are accepted

Phase 1 art drops from 103 to roughly 88, the plush lane starts earlier at the same total spend, and the riskiest assumption in the business case — that a recolor is cheap — gets tested first instead of at sculpt 18.

WootBuild will build whichever plan docade chooses.