@mattshumer_: For all of you who think this is fake... here's the repo and prompt (in the thread below) that generated this in one sh…

X AI KOLs Following Tools
ai-generated three-js first-person-shooter procedural-generation webgl open-source game-development

Summary

Matt Shumer shares the GitHub repo for 'Claude of Duty', a first-person shooter built entirely in the browser using Three.js and WebGL2, with all assets procedurally generated by AI agents at load time.

For all of you who think this is fake... here's the repo and prompt (in the thread below) that generated this in one shot: https://github.com/mshumer/Claude-of-Duty…
Original Article
View Cached Full Text

Cached at: 07/25/26, 10:12 PM

For all of you who think this is fake… here’s the repo and prompt (in the thread below) that generated this in one shot: https://github.com/mshumer/Claude-of-Duty…

mshumer/Claude-of-Duty

Source: https://github.com/mshumer/Claude-of-Duty

Claude of Duty

Get updates here.

A first-person shooter built in the browser with Three.js r180 and WebGL2. Roughly 55k lines across 11 subsystems, written by a fleet of AI agents under orchestration.

There are no art assets. Every texture, mesh, animation and sound is generated procedurally at load time from code. No models, no HDRIs, no image files, no audio files. The only runtime dependency is three.

npm install
npm run dev          # http://127.0.0.1:5173

Click the canvas to lock the cursor. WASD move, mouse aim, LMB fire, RMB ADS, R reload, Shift sprint, Ctrl crouch, Space jump, Q/E lean, Esc release.

What’s in it

subsystemwhat it does
renderHDR pipeline, cascaded shadow maps in a sampler2DArray with texel snapping and PCSS contact hardening, MRT depth/normal/velocity prepass, GTAO, TAA with YCoCg variance clipping, tile-dilated motion blur, Karis bloom pyramid, GPU EV100 metering, procedural 33³ grade LUT, AgX composite
materialsGPU texture forge: 19 procedural surfaces (concrete, brick, plaster, asphalt, sand, rusted/painted/brushed metal, wood, fabric, burlap, glass…), periodic noise so everything tiles seamlessly, Sobel height→normal, parallax occlusion mapping, triplanar projection, curvature-driven edge wear
skyAtmospheric scattering, time of day, PMREM environment generation, volumetric fog and light shafts
world~120×120 m market street: modular building kit with real wall thickness, enterable interiors, several hundred instanced props
physicsWritten from scratch, no library. Binned-SAH BVH (29k tris → 14k nodes in 22 ms, 0.25 µs/raycast), swept-capsule character controller with a 5-plane crease stack, impulse rigid bodies with CCD, PBD ragdolls, multi-layer bullet penetration
playerMovement state machine, slide/mantle/lean, camera feel
weaponsProcedural weapon geometry, viewmodel rig, ADS, spring recoil, procedural reloads, ballistics with travel time and drop
fxGPU particles, decals, tracers, muzzle flash, explosions
aiSkinned soldiers, navmesh pathing, perception, cover behaviour, ragdoll death
uiDOM/CSS HUD: crosshair, hitmarkers, minimap, compass, killfeed
audioWeb Audio synthesis — no sound files. Layered weapon fire, convolution reverb, HRTF spatialisation, occlusion

ARCHITECTURE.md is the contract the agents worked against: subsystem interface, directory ownership, the cross-subsystem event vocabulary, and shared surface types.

Tooling

The interesting part of this repo is arguably the harness, not the game.

toolpurpose
tools/capture.mjsScreenshot one named shot via GPU-backed headless Chromium
tools/shotset.mjsAll 11 shots in one session — fast review set
tools/baseline.mjsReproducible capture: each shot in an isolated page, fixed frame budget. Bit-identical across runs
tools/imagediff.mjsPer-pixel gate. Exits non-zero if any pixel moved
tools/profile.mjsGameplay profiler at real device pixel ratio. Frame-time distribution and hitch attribution via per-frame WebGL program counts
tools/playtest.mjsScripted movement/fire smoke test

Two findings worth recording, because both invalidated earlier measurements:

Median frame time hides the actual problem. A static-camera benchmark reported 94 fps while the game was unplayable. Real gameplay at Retina DPR (internal 3.34 MP, not 2.07) ran 12–17 fps with 728–1236 ms stalls caused by 34+ WebGL programs compiling lazily mid-frame. profile.mjs reports p50/p95/p99 and attributes each hitch, which is what surfaced it.

Captures were not reproducible. shotset.mjs reuses one page across all 11 shots, so particle age, decal buffers and exposure state leak forward — two identical runs differed on 10 of 11 shots. baseline.mjs isolates each shot in a fresh page, which is bit-identical and is what makes imagediff.mjs a usable gate.

Performance

Measured on an Apple silicon laptop at 1512×982, DPR 2 (3.34 MP internal), ultra preset, 3 runs, gameplay in motion with AI and firing active:

before optimizationafter
fps p5012–1728–30
fps p994–914–17
worst frame728–1236 ms66–82 ms
shader compiles during play34–350
boot~9–12 s3.7–4.6 s

The optimization pass was constrained to produce zero visual change, enforced by imagediff.mjs rather than by assertion — the shipped build is bit-identical to its pre-optimization reference across all 11 shots.

Shader pre-warm (src/core/prewarm.js) is what removed the stalls. Making it provably pixel-neutral required first fixing subsystems that animated off performance.now() instead of the engine clock, since any change to boot duration otherwise shifted output.

Honest assessment

The goal was to match a modern Call of Duty. It does not.

Eleven independent adversarial critics scored the frames against that bar. Scores went 3.59 → 4.14 → 4.05 → 5.05 out of 10. Two shots reached “CLOSE”; the rest remain “AMATEUR”. In a blind A/B, every critic in every round picked the real Call of Duty frame.

Where it falls short, specifically:

  • Hands. Blocky finger slabs that don’t convincingly grip the weapon.
  • Material richness. Surfaces read as procedural noise rather than photographed reality at close range — the ceiling of generating texture from code.
  • Characters. Enemies read as mannequins at distance.
  • Indirect light. An approximation, not real GI.
  • Frame rate. 28–30 fps at Retina. The art passes tripled geometry cost (5.9M → 11.3M triangles) and optimization recovered about half.

A known root cause remains unfixed: the viewmodel light rig in render/index.js delivers roughly 20× the irradiance per unit albedo that the world does — a plain black material in the view scene renders at L=110 against a background of 91, purely from F0=0.04. Every weapon albedo is cheated to a third of physical to compensate, which caps material separation on the most-looked-at object in the game.

Process note

Sequential single-owner passes beat parallel fan-out decisively. Three rounds of six agents each owning one directory moved the score +0.46 and left frame-ruining defects higher than they started (60 → 47 → 66), because tonemapping, sky and indirect light are one coupled system and isolated agents kept breaking each other’s assumptions. One sequential pass with a single owner per coupled concern moved it +1.00 and cut defects 66 → 26.

The most valuable single result came from an agent contradicting its own brief. Every critic for three rounds reported the weapon as “untextured”. It wasn’t — it was specular-dominated, with the diffuse term measured at L=26 against a shipped L=67. Prior rounds had been crushing albedos to fight bright-part complaints, which killed diffuse and made it worse. The fix was the opposite of what was asked for.

Matt Shumer (@mattshumer_): Claude Opus 5 one-shotted this game.

EVERYTHING you see in this demo is custom code… not a single external asset was used.

AI games are going to be amazing.

(sound on)

Similar Articles