How the world gets its next chapter¶
Status: PvE campaign delivery plan · Revised 8 September 2026
Fight the Frontier's advantage should be the time between a meaningful player action and a well-made response. AI can help us author quests, creatures, environments, dialogue, tools and tests quickly. The player-facing promise is a responsive world with coherent consequences.
Two different kinds of change¶
Live simulation uses shipped, tested rules: patrols move, supplies run low, factions raid, an objective changes hands. The authoritative server owns those state transitions.
New content changes the authored game: another dungeon, a quest branch, new equipment, a region or an enemy. AI assists its creation; it enters through a reviewed content pack. Live player messages and faction events do not execute generated code or bypass release checks.
A small repeatable pack¶
| Step | Output |
|---|---|
| Observe | Durable event result, player behaviour and specific feedback |
| Brief | One clear purpose, selected branch, affected places and the benefit to players |
| Approve | Alex reviews the concrete scope, rewards, consequences and exclusions |
| Author | Data, original Blender assets, encounter routes, dialogue and quest UX |
| Validate | Schema/reference checks, migrations, repeatability, visual inspection, performance and multiplayer journeys |
| Review | A running preview, concise change record and evidence for the exact pack |
| Activate | Versioned content manifest and compatible saves; explicit release decision |
| Learn | World chronicle, player response, defects and measured authoring/verification time |
Pack manifest¶
Each pack records its ID and version, required runtime/data versions, event branch, quest/object/creature IDs, asset hashes, migrations, reward policy, compatibility and rollback. Keep approved scope with the pack. Existing objective IDs retain meaning; migration handles any deliberate change.
Changes to rewards, ownership and inventory require idempotent server transactions. Read-only UI text and visual assets can take a lighter verification path. Faster delivery comes from reusable validated templates, automated reference checks and small scope, with stronger checks where state can be lost or duplicated.
The first proof of rapid development¶
Use M1 to measure the complete cycle for a quest/creature pack. Use M3 to rehearse persistent war outcomes, then M7–M8 to prove that a recorded six-frontier campaign outcome can select, verify and activate a corresponding pack. Track authoring time, review time, defects, rollback success and player completion. Set a sustainable release cadence from those results.
Player feedback can propose destinations, enemies and conflicts. World-event results select among clear promised consequences. Alex approves what is built and released; this wiki records decisions so the next fast iteration stays consistent with the world.