# Quick start: play your first world

LifePot is a bounded, replayable artificial-life game—not a biological forecast. You can play without installing anything or supplying an API key.

## Open the game

From the [LifePot overview](/), choose **Explore this example** to enter the recorded pond immediately, with no setup questions or new inference. The landing preview is real recorded gameplay, not a live simulation. Its video stays unloaded under reduced-motion or data-saving preferences until you choose Play preview. Inside the example, optional tips explain pausing, inspection and family following; you can dismiss them.

Open [/play](/play) to preview three authored ecologies: **Resource competition**, **Hunters and prey**, and **Recurring scarcity**. The square habitat preview shows the selected starter's actual generation-0 founders and resource field. These are deliberate starting configurations, not Jev interpretations or guaranteed outcomes.

Select **Review this world** to open its editable conditions. Starters default to **Fixed world rules**, so previewing, reviewing and playing them makes no inference requests. **Seed ecosystem** starts the observation only after review. Changing the policy to adaptive explicitly enables the existing runtime council path.

Choose **Describe your own world** for the three-question creation path. Review the interpretation and its provenance before seeding. An unresolved or failed interpretation remains blocked rather than being silently approved. **Explore deterministic preset** remains available as a separate reviewable configuration.

Review the actual resources, hazard, species roles and food web before starting. Resource competition has two resource-feeding species without a predation link; hunters and prey has a directed feeding link; recurring scarcity uses scarce resources and pulsed drought settings. Initial conditions do not guarantee survival, and these starters change several conditions rather than representing a one-condition experiment.

## Observe a generation

New worlds start at **Slow**, targeting one generation per second; the landing example starts at **Normal**. Use the **Speed** selector to choose Normal or Fast when you want to move through the run more quickly. Speed changes timing only; it does not change the world rules or replayed result.

Watch population, resources and the generation counter. Pause before selecting a living organism in the Creature Inspector. Follow its family to examine lineage context, or use the World Observatory for aggregate measurements. Resume to watch what changes. Selection observes the world; it does not change its rules.

Local simulation rules govern movement, feeding, reproduction, mutation and death. In adaptive worlds, Jev can respond to the ecology with validated environmental changes, species strategies and mutation policies, not direct control of individual organisms. A fallback or abstention is not a live model decision. The run ends at extinction or its observation horizon; completion is not proof that your idea succeeded.

## Try a second experiment

Read [observations and outcomes](/learn/observe-and-inspect), then [experiments and comparisons](/learn/experiments). Review a fixed-rule baseline when you want to hold the runtime policy steady; changing multiple conditions or making fresh adaptive decisions is exploratory, not a controlled causal claim.

Use a completed run's challenge/replay controls to inspect recorded behavior. [Replay](/learn/replay) reproduces recorded rules and choices without new interpretation; it does not establish biological accuracy or guarantee compatibility with every historical engine.

## Run locally

Clone the [repository](https://github.com/tugrulguner/lifepot), enter its root, and use its locked dependencies:

```sh
npm ci
npm run dev
```

Open the localhost URL printed by Next.js, then visit `/play`. A server-side model key is optional; absent credentials use the game's labeled fallback paths. Never put provider credentials in browser code. [Deployment and protection](/learn/deployment) explains the Cloudflare build and inference boundaries.

Continue with [setup and review](/learn/setup-and-review), [developer contracts](/learn/contracts), or the [deterministic engine internals](/learn/engine-internals).