Velmora

Velmora is a skill-based melee PvP game set inside a persistent social hub city, built for Roblox, and the setting underneath it is more deliberate than "generic fantasy world": it's a real fusion I built on purpose, merging Marvel's Asgard with a separate gothic-city concept I'd been developing called Velmora City, into one continuous fictional empire I call the Eternal Imperium. The sun above it is permanently bound in a "Blood Eclipse," known in-world as the Crimson Eye, and the entire palette is locked to exactly three colors (gold, crimson, grey-black stone), because a world that commits to a narrow palette reads as designed, and one that doesn't just reads as assembled from whatever asset pack was open that day.

The core loop is server-authoritative melee combat: twenty-one handcrafted weapons across eight classes, from scythes to greatswords to flails, eleven server-side abilities, every one of them telegraphed so a hit never feels like it came from nowhere. Duels aim for somewhere around eight clean hits to kill and thirty to seventy-five seconds per fight, which is a deliberately narrow target: fast enough that losing doesn't feel like a wasted ten minutes, slow enough that skill actually shows up in the outcome instead of the first person to land a lucky hit winning by default.

The rule I hold hardest

Progression in Velmora buys identity, never power. Five Covens exist purely as social/status factions; no stat bonuses attached to any of them. The battle pass, Eclipse Pass, is cosmetics-only. This is written into the project's own constitution file as one of five Iron Rules, and it's the one I'd fight hardest to protect if anything ever pressured me to compromise it, because pay-to-win is the single fastest way to turn a skill-based game into a game about whoever spent the most money instead of whoever's actually good. I'd rather have a smaller, fairer game than a bigger, resentful one.

The system I actually built the game around, more than the combat itself, is the Eternal Ledger, a nemesis and rivalry tracking system that remembers who's beaten you, who you've beaten, and turns that history into something the game surfaces back to you. Most PvP games treat every match as disconnected from the last. The Ledger is my answer to the question of why anyone should care about an opponent beyond the sixty seconds of a single duel: because the game remembers, and so, eventually, do you.

Under a month, near-daily, and it shows in the numbers

Two hundred and fourteen Luau script files. Three hundred and twenty commits. The entire documented history runs from mid-June to early July, under a month of intensive, close-to-daily solo work, backed by a checkpoint daemon running every ten minutes via Windows Task Scheduler, silently autosaving recovery breadcrumbs in case something crashed mid-session. The internal documentation is its own kind of artifact: a running build log over thirteen hundred lines long, a module-contracts document pushing two thousand, and a lore bible (the Imperial Velmora-Asgard Reconstruction document) that runs to over eleven hundred lines and nearly two hundred kilobytes, written in full encyclopedic reference style: cosmology, geography, economy, military history, a production bible for a game world that most solo Roblox projects would never bother documenting this thoroughly.

I run this like it has an engineering org even though it's one person. A constitution file lays out five Iron Rules governing every decision: server authority over the client, no pay-to-win ever, one write pipeline for state changes, integrated quality over shipping broken and fixing later, contracts written before code. That's not me performing professionalism for an audience. It's the actual discipline that kept two hundred-plus script files from turning into an unmaintainable mess over a month of near-daily changes with nobody else reviewing my work before it shipped.

The rig that fooled me for weeks

Here's a mistake I'll own plainly: I believed for weeks that the character rig used some kind of broken custom constraint system, and I blamed a string of animation problems on that assumption, working around a rig I thought was fundamentally unusual. Late in June I discovered it was the completely standard Roblox R15 rig the entire time. Nothing custom, nothing broken, just my own incorrect model of what I was working with, compounding into weeks of wasted debugging built on a wrong premise. That's the kind of mistake that's genuinely embarrassing in the moment and genuinely useful afterward, because it recalibrated how much I trust my own assumptions about a system before I've actually verified them against the platform's real behavior instead of what I expected the behavior to be.

I also rejected my own first pass at the combat system after playtesting it myself: every weapon shared the same animation regardless of class, and it felt worse than Roblox's own default R6 combat, which is about as unflattering a comparison as I could have made against my own work. That led to an eight-workstream overhaul: real per-weapon animation, a proper aim camera, a full VFX and sound presentation layer, physics-driven cosmetic motion like cape sway, a spring-mass-damper system for pet followers that actually reads as alive instead of rigidly snapping to position. I'd rather scrap a system I built and rebuild it properly than ship something I know, from playing it myself, doesn't feel right.

Where the "built and playable" claim needs a footnote

My site describes Velmora as built and playable, and the first milestone genuinely is: combat, movement, abilities, the Ledger, ranked play, covens, the core social hub all work. What the site doesn't say, and what I'll say here, is that the game's economic backend is still unfinished even post-launch: the dev-product IDs for coin bundles and the premium pass tier are placeholders, not connected to real Robux transactions yet, and Roblox's own analytics dashboard wasn't even enabled at the point of the last commit I have record of. Trading was fully built and then deliberately cut and locked, over real, specific fears about duplication exploits, scams, and real-money trading risk: a product decision I made on purpose, not a feature I ran out of time for. And several animation and sound assets are honestly rougher stand-ins right now, waiting on a proper pass once the systems underneath them stop shifting.

Full stack developer. Founder of Yashveer Labs. A month of near-daily work, one Iron Rule I won't break, and a game world I took seriously enough to write its own encyclopedia before I finished building its combat.