Animal Lineage Runner Project Report
1. The Concept
Animal Lineage Runner is a mobile-first endless runner where players begin as a small rabbit and automatically evolve into stronger animals as they survive longer. Instead of unlocking characters through currency or premium purchases, every evolution is earned purely through gameplay and distance milestones. Each animal introduces a unique movement style and special ability, creating a sense of progression within a single run while keeping the controls simple and accessible.
I chose this project as a one-day sprint because endless runners appear deceptively simple but require many interconnected gameplay systems working together in real time. It became an excellent test of whether an AI planning workflow could design a scalable game architecture and then hand those requirements to another AI for implementation using only HTML, CSS, and JavaScript.
---
2. The Stack
The Stack: HTML5, Vanilla CSS, SVG Assets, and Vanilla JavaScript.
The AI Assist: ChatGPT handled the entire product planning process, gameplay architecture, balancing, UI/UX planning, animation specifications, asset guidelines, and technical handoff documentation. OpenAI Codex was then used for the implementation phase, translating the markdown specifications into a fully playable browser game. Throughout development, multiple prompt iterations were required to refine gameplay, improve visuals, fix animation issues, and optimize performance.
---
3. The 1 Day Sprint
Morning (Hours 1–3): The project began with planning the overall gameplay loop, progression system, and technical architecture. Instead of building a traditional endless runner, we designed an evolution mechanic where the player transforms into different animals during a single run. We carefully defined evolution thresholds, abilities, responsive layout requirements, animation states, and mobile controls before producing a structured markdown handoff document for Codex.
Afternoon (Hours 4–6): The implementation phase focused on building the core gameplay systems. Codex successfully generated the runner engine, obstacle spawning, collision detection, scoring system, evolution logic, and animal abilities. However, many systems required repeated prompt refinement. The generated UI initially resembled a prototype, animations lacked polish, character proportions were inconsistent, and several gameplay interactions behaved unexpectedly. Considerable time was spent iterating on animation states, sprite generation, responsive layouts, HUD organization, environment transitions, and performance optimization.
The Finish Line (Hours 7–8): The final hours focused entirely on polish rather than adding new mechanics. We improved the responsive layout, refined the HUD, optimized SVG assets, added environmental transitions, enhanced particle effects, balanced abilities, optimized rendering performance, and completed multiple QA passes. While the game became fully playable with smooth progression and multiple animal evolutions, the visual presentation still required additional art refinement to reach commercial-quality standards.
---
4. The Roadblocks
Roadblocks: The biggest challenge was maintaining consistency between AI-generated planning and implementation. While ChatGPT produced extremely detailed gameplay specifications, Codex occasionally interpreted those instructions differently, especially regarding animation quality, UI composition, sprite generation, and visual polish.
Another major challenge involved sprite creation. Although Codex could generate functional SVG assets, many early character sprites contained distorted anatomy, disconnected ears, inconsistent proportions, cropped limbs, or unnatural animation cycles. Significant manual prompting and repeated regeneration were required before the animals began resembling proper game characters.
Balancing gameplay systems also proved more time-consuming than expected. Features such as evolution timing, unique abilities, obstacle spacing, cooldown management, responsive scaling, collision handling, and environment transitions all interacted with one another, requiring repeated testing and refinement to maintain a smooth gameplay experience.
---
5. Key Takeaways
**What worked the best?
** The planning phase was the strongest part of the entire workflow. ChatGPT generated a detailed gameplay architecture, progression roadmap, technical specifications, animation requirements, responsive layout guidelines, and optimization strategy within minutes. Having a structured development roadmap significantly reduced implementation uncertainty and kept the project organized despite its complexity.
**What worked the worst?
** Visual interpretation remained the weakest part of the multi-AI workflow. While Codex performed well implementing gameplay systems, it frequently struggled to match the intended artistic direction. Character anatomy, animation quality, sprite consistency, HUD aesthetics, and environment composition required extensive manual prompt iteration before approaching the desired quality level.
My final thought: If I had one more day, I would focus entirely on presentation rather than mechanics. The gameplay foundation is already complete, so the next milestone would be replacing placeholder SVG artwork with production-quality assets, improving character animation through smoother frame interpolation, adding richer environmental effects, enhancing audio, and refining the overall game feel. If I were to repeat this project, I would still use AI-assisted planning, but I would divide development into much smaller implementation stages with frequent validation after each milestone. That approach would reduce misinterpretation, minimize rework, and produce a more polished final game within the same development time.