Inspect a creature in the ecosystem, export its genome or experienced state, then add it here. You can also import an entire population.
Shared resources, inherited controllers, no assigned winner.
Import a genome, experienced individual or population. Source identities and ancestry are retained; arena IDs and placement are remapped.
No file selected
Experienced copies preserve the same past, then diverge as independent bodies. Population import uses one copy. Competition is ecological, with no scripted strategies.
Inherited neural controllers choose movement, feeding, attacks, reproduction and chemical deposition. No hunts, alliances, victories or breakthrough events are scripted.
Every displayed body is driven by real physical genes. This first version uses a continuous single-body anatomy: circular collision body, energy reservoir, eyes and propulsion/attack glyphs. It does not evolve articulated limbs or new organ topology.
DNA acquisition requires actual animal biomass transfer, genetic compatibility, an uptake cost, a 120-tick metabolic integration and a 600-tick cooldown. One physical gene is blended per fragment. Completed changes enter the reproductive genome; unfinished integration and learned offsets are not inherited.
Memory is the inherited controller’s 10 recurrent units. Optional uncorrected SR learning uses energy/health feedback. Synthetic competence exists, but general ecological usefulness has not been demonstrated. The rejected clipped-score correction is not in use.
The arena ledger describes current living competitors and their lifetime totals, including imported histories. It is not an unbiased tournament score. Genome/individual imports retain qualified ancestry; their local IDs and spatial coordinates are remapped. A population file also retains the source world, resources and RNG; arena insertion merges its living bodies into shared resources.
All worlds are isolated from production. Research classification remains NO_PRACTICAL_CORRECTION_BENEFIT and production learning remains OFF.