Standout Builder Studio · planning Index / Journey simulator review Updated 13 Jun 2026

Play the journey

Pick a build, click through all eight steps, and hit the same gates a real builder would. This is the proposed guided flow made tactile — a reskin that breezes past the math gate, a tune that gets stopped cold by it, an import that forks on a mechanic mismatch, and a brand-new game that waits on the one engineering review. The timers and agent outputs are scripted; the step order, gate logic, agent roster, and config values are the real proposal.

Choose your build

Four scenarios, four shapes of the same flow. Each one bites at a different step — that is the whole point of having a path picker. Pick one to start a run.

The simulator needs JavaScript
Without it, these cards link to the matching static worked-example walkthroughs instead, and the compare table below is fully readable on its own. With JavaScript on, clicking a card starts an interactive run in place.
Reskin ~2:05 to playable
Neon Heist
A cyber-noir skin of the certified Crash mechanic. New palette, art, and copy — zero config change.
Bites at: step 4 (assets) — a lobby tile fails the mobile safe-area check. No math gate.
Start this run →Read the walkthrough →
Tune ~3:10 to playable
Trunk Raider: Overdrive
A higher-volatility cut of Trunk Raider. Four config knobs pushed toward more extreme outcomes — and a math owner who must sign off.
Bites at: step 5 (validation) — the RTP sim and the human math gate. Pick the wrong preset and it fails.
Start this run →Read the walkthrough →
Import 0:25 to spec
Golden Galleon
A third-party pick'em arrives as three screenshots and an info-sheet PDF. Extract, match, and decide: adapt or flag a gap.
Bites at: the pre-step fork — match is 87%, so you choose between a variant or a respectful mechanic-gap dead-end.
Start this run →Read the walkthrough →
New game ~days, compressed
From a mechanic
A brand-new game built from a library mechanic. Traverses every step — and hits the one structural engineering review.
Bites at: a Scaffolder PR after step 5 that an engineer must merge before a sandbox can deploy.
Start this run →Read the deep-dive →
CONCEPT · proposed elapsed0:00
Behind the scenes
Agent activity appears here as the run progresses.
Gate states

Compare the paths

The same eight steps, four very different journeys. The heavy step and the human approvals are what separate an hours-scale path from a day-scale one.

Scenario Heavy step Gates hit Human approvals Time-to-playable
Reskin · Neon Heist you played this Step 4 · assets (safe-area) Theme, art, audio, copy — no math gate 1 · art owner approves the regenerated tile 2:05
Tune · Trunk Raider: Overdrive you played this Step 5 · validation (RTP sim + math gate) Manifest, math config, RTP sim proof, math-owner approval, assets, jurisdiction 1 · math owner approves rtp-proof #341 3:10 (+0:40 on a Redline retune)
Import · Golden Galleon you played this Pre-step · extraction + match fork Extraction confidence, mechanic match (87%), then the standard variant gates incl. one math gate 1–2 · confirm low-confidence fields; math owner if variant chosen 0:25 to spec, then ~3:00 to playable
New game · from a mechanic you played this Every step + a Scaffolder PR gate before step 6 All gates, plus the one structural engineer review on the scaffold PR 2+ · engineer merge, math owner, art owner ~days (compressed for the simulator)

What this fakes — and what it doesn't

Scripted
What the simulator fakes
  • The clock is a mock counter — it jumps to scripted beats, it does not measure real work.
  • Agent outputs (sim results, deploy transcripts, extraction confidences) are pre-written, not computed.
  • Launch URLs and session rows are illustrative shapes; nothing is signed and no Worker is deployed.
  • Nothing persists — reload the page and every run is gone. No data leaves the page.
Faithful
What it reproduces
  • The canonical eight-step order and which steps each path skips or locks.
  • The gate logic: a reskin has no math gate; a tune cannot pass until the math owner approves; a failed RTP retunes.
  • The agent roster — Scaffolder, Math-Sim, Asset, Copy/Localisation, Validation, Deploy, Cert-Pack, Info-Sheet — and where each one acts.
  • Real config values: the 9 Crash fields, the 4 Overdrive deltas, the RTP figures (95.97% / 94.81%), and the import confidences. See the generative AI map.