One prompt lit the forge.

The spark

The impossible arcade cabinet

“Multiplayer online… hosted on a VPS. Four-player co-op.”

Emberbound did not begin as a planned production. It began as the same loose challenge handed to Claude Code, Grok Build, and Codex. The first prompt supplied energy, not a specification; Codex supplied the strongest playable answer, and Jeff chose to keep pushing it. The online and VPS follow-ups immediately ruled out a disposable single-player mockup.

The growing wish list used genre landmarks as emotional shorthand: Golden Axe for fantasy-brawler impact, Gauntlet: Dark Legacy for long-form cooperative adventure, the four-player Teenage Mutant Ninja Turtles and X-Men cabinets for shared-screen clarity, Street Fighter bonus rounds for joyful detours, and Diablo for online persistence. They defined feelings to chase—not characters, worlds, or assets to copy.

What it became: an original browser-native realm with one authoritative Node.js simulation, private four-player rooms, and WebSocket play.

The quality line

“The dude’s not even holding the weapon.”

“It’s still very amateur.”

That blunt review established the project’s most useful rule: technically present is not visually convincing. Weapons needed to belong in hands. Feet needed to meet the floor. Walkable ground, camera movement, collisions, approach distance, and enemy entrances all had to tell the same physical story.

What it changed: grounded presentation, stronger hero renders, authored spawn approaches, readable combat lanes, clearer hit geometry, and a much higher acceptance bar.

Combat identity

Every attack needed a body behind it

“There’s no body movement… it just stands there.”

Sable’s projectile could not simply appear at a weapon tip. Casters, archers, bruisers, and guardians needed different windups, release points, recoil, combo branches, and recovery poses. Secondary attacks grew into character-specific motions; light, heavy, ranged, and melee inputs became combo vocabulary.

What it changed: six combat identities, visible equipped weapons, ranged casting origins, multi-hit sequences, character death performances, boss finishers, and rotating voice reactions.

Fight as one

Co-op became more than four people attacking nearby

“Partner abilities… unique as to who initiates.”

The warband needed verbs that only existed because another player was present: linked spins, throws, shared specials, rune relays, rescue pressure, and score rivalry inside cooperative bonus trials. “T for team up” eventually became both the signature partner input and the readable way to resurrect a downed ally.

What it changed: directional team attacks, downed timers, revive signaling, full-party HUD information, cooperative challenges, and shared victory moments.

Arcade becomes campaign

The fight gained memory

“Play for 20–30 minutes at a time and return together.”

That request transformed a run-based brawler into a persistent online adventure. Accounts, heroes, campaigns, weapons, unlocks, difficulty, checkpoints, reconnect grace, and safe return points now mattered as much as moment-to-moment combat.

What it changed: durable PostgreSQL-backed chronicles, six selectable heroes, equipment progression, Novice/Veteran/Hardcore modes, twelve stages, and resumable shared campaigns.

A living warband

The lobby became a social place

“Friends… individual chat, party chat, voice.”

Real cooperative play needed relationships outside the battlefield and coordination inside it. Authentication, email verification, friend requests, direct messaging, party chat, presence, and opt-in WebRTC voice became platform systems rather than combat shortcuts.

What it changed: persistent identity, social rosters, private and party channels, mute/deafen controls, and TURN-ready voice architecture.

The pivotal question

Was this a game—or the beginning of an engine?

“I want to create additional games like this… built solid, and replicatable.”

The answer could not be a renamed folder full of Emberbound assumptions. The reusable pieces had to stop importing the game. Time, random state, transport, content, and game rules needed explicit boundaries. A second headless game had to install the packed packages outside the workspace and prove they worked without Emberbound.

What it changed: five private @generated-adventures/* packages, strict dependency boundaries, a clean-room Smoke Arena consumer, and documentation for creating another game.

The rebuild

Reliability became a feature players should never notice

“Back up what we have… make sure it’s reversible and test it in the new engine.”

The migration could not gamble with a working game. Authority moved behind versioned protocols and fixed simulation steps. Reconnect behavior, deterministic state, checkpoint envelopes, recovery artifacts, journaling, storage boundaries, and rollback were tested as hostile failure surfaces—not optimistic happy paths.

What it changed: engine-owned room authority, deterministic replay foundations, bounded recovery evidence, immutable release artifacts, automated rollback rehearsals, and canonical full-campaign raids before deployment.

From prototype to public Alpha

The realm acquired a front door

“The Alpha on EmberboundGame.com… gate who gets access.”

The game, the public story, and player acquisition became separate surfaces. The root domain could sell the dream without consuming realm resources. The play subdomain could protect accounts and campaigns. Alpha applications could enter a consent-aware recruitment workflow without exposing invitation security.

What it changed: a cinematic marketing site, protected application intake, branded transactional mail, HighLevel automation, a locked-down VPS, verified deployments, and a private GitHub release trail.

The journal opens

The build process becomes part of the game’s story

“A ‘Codex Journal’ section… show how a simple prompt turned into this massive project.”

The latest prompt makes the collaboration visible. Not as magic and not as a transcript dump, but as evidence that strong direction, honest testing, and the willingness to rebuild weak foundations can turn a loose idea into a serious living system.

What it changes: players can now follow the decisions behind the Vale—and watch future entries arrive as Alpha feedback shapes what comes next.