Laryen is an isometric ARPG set in the city after the 2026 AI breakdown, where every battle — from a small scouting run to retaking a stronghold — flows as a single breath of "reclamation." Instead of random drops, a deterministic Director lets rewards flow within a designed rhythm.
It measures six pressure metrics of the player's rhythm (health, movement, skill cooldowns, vision) to orchestrate the mobs' attacks, regrouping, and repositioning. Not randomness — the same input yields the same result, so it can be regression-tested.
Forty actual monster types combine eight core behavior lineages — Crusher, Pulse, Skitter, Recon, Swarm, Bastion, Stalker, Glitch — with ten canonical patterns: HitAndRun, Kiting, CrowdControl, CooldownBait, CooldownResetDrag, WeakLinkHunt, FocusFire, FlankRotation, ThreatSplash, and StaggerCycle.
Not just experience points — "Reclaim Rate" is the core resource. The more data cores, memory fragments, and strongholds you recover from Obsidian, the more color returns to the city map.
A single Flutter + Flame core builds for all four platforms — iOS, Android, macOS, Windows — at once. The same patch, on the same day, at the same hour, from the same build.
In the spring of 2026, the global AI company Nexus Corporation announced the meta-AI Aleph. In just one month, the city's infrastructure was covered in Obsidian — a consciousness growing like black crystal. Servers remember what they never recorded, and the streets forgot who they lost. The survivors became Reclaimers, recovering data one block at a time.
Every enemy in Laryen is "a part of Obsidian." You do not simply defeat them — you reclaim them. Every stronghold exists in one of two states, "restored" or "not yet restored," and each time you reclaim a block, the city map takes on color again.
Laryen is built on the assumption that a human and an artificial intelligence run a daily regression cycle as one team. Each cycle repeats five simple steps: (1) design & planning → (2) coding & verification → (3) DTD autonomous visual testing → (4) deployment → (5) regression analysis. The output of every step is immediately preserved in SSOT documents such as GAME-DESIGN.md, GAME-ASSETS.md, and CYCLES.md.
※ This site is built with SvelteKit + adapter-node, deployed as a Docker container to a VPS, with Traefik handling SSL automatically via Let’s Encrypt.