Rekindle
Six disasters. One fire. Rebuild civilization.
▶ Watch video walkthrough
Outcome
A coherent journey from survival to rebuilding society.
- ✓Distinct disaster worlds share consistent core rules
- ✓Learning is paired with illustrated methods and sources
- ✓Progress and multiplayer changes are checked before saving
- The challenge
- A deep survival simulation needs coherent consequences, useful learning and progression that stays consistent across very different worlds.
- How I led it
- Game and product design, sourced learning content and artificial intelligence (AI)-assisted development of survival rules, progression and multiplayer systems.
- What I delivered
- Rekindle combines fixed survival and society rules with sourced journal content, six disaster packs and bounded AI evaluation of player ideas.
Deep dive
Open any section for detailHow it worksFrom first shelter to society
- 01
Choose
Select a disaster world
- 02
Survive
Meet basic needs
- 03
Learn
Read sourced journal methods
- 04
Invent
Develop tools and settlements
- 05
Rebuild
Trade and make civic choices
Challenge and solution, side by side2 examples
- ChallengeAn AI idea could bypass the economy.SolutionThe idea judge evaluates a proposal, but fixed rules cap the outcome.
- ChallengeReturning after a pause unexpectedly advances the camp.SolutionPause rules freeze catch-up and changes; dedicated source tests cover this behavior.
Product focus and qualityFocus, validation approach and key features
Product focus
Combine an illustrated field journal with a rule-driven world where survival, invention and community choices have consequences.
Quality & validation
Source tests span survival, weather, economy, progression, multiplayer, pause behavior and save handling. A shared World Standard checks consistency across disaster packs. Core outcomes follow fixed rules; the optional AI idea judge operates within those limits and does not generate survival facts.
Key features
- ·Six disaster worlds with distinct conditions and progression
- ·Illustrated skill journal with methods, warnings and sources
- ·Body, shelter, food, water and weather systems
- ·Invention, workshop and settlement development
- ·Tribes, trade, clans and civic decisions
- ·Optional AI assessment of player ideas within fixed rules
- ·Revision-checked saves and pause handling
- ·Completed-journey books and printable help
Why it matters
An example of making a complex, interconnected product testable without letting AI improvise its underlying rules.
Scope, notable components and compliance
Measurable scope
- ·Six disaster packs
- ·Source regression suites span survival, trade, society, saves and world consistency
Inside the app and demo videos3 screens · 1 demo video
Demo videos
Tech stack and skills
Tech stack
TanStack StartReact and TypeScriptLovable CloudPure TypeScript game rulesBounded AI idea evaluationVitest
Skills demonstrated
Systems modelingGame designRule-based AI boundariesRegression testing
▤ README.md›
# Rekindle > Six disasters. One fire. Rebuild civilization. A painted survival-journal game spanning six disaster worlds, from the first shelter to rebuilding a society. ## Outcome **A coherent journey from survival to rebuilding society.** - Distinct disaster worlds share consistent core rules - Learning is paired with illustrated methods and sources - Progress and multiplayer changes are checked before saving ## Problem A deep survival simulation needs coherent consequences, useful learning and progression that stays consistent across very different worlds. ## Solution Rekindle combines fixed survival and society rules with sourced journal content, six disaster packs and bounded artificial intelligence (AI) evaluation of player ideas. | Problem | Solution | | --- | --- | | An AI idea could bypass the economy. | The idea judge evaluates a proposal, but fixed rules cap the outcome. | | Returning after a pause unexpectedly advances the camp. | Pause rules freeze catch-up and changes; dedicated source tests cover this behavior. | ## Features - Six disaster worlds with distinct conditions and progression - Illustrated skill journal with methods, warnings and sources - Body, shelter, food, water and weather systems - Invention, workshop and settlement development - Tribes, trade, clans and civic decisions - Optional AI assessment of player ideas within fixed rules - Revision-checked saves and pause handling - Completed-journey books and printable help ## Tech stack - TanStack Start - React and TypeScript - Lovable Cloud - Pure TypeScript game rules - Bounded AI idea evaluation - Vitest ## Scope - Six disaster packs - Source regression suites span survival, trade, society, saves and world consistency ## Links - Live: available on request ## Author Candace Hagood · Product & Quality Leader · AI-Assisted Engineering