threejs-testing — Claude Code skill: what it does, how to install
threejs-testing — a skill for Claude Code defines visual QA and browser testing guidelines for Three.js scenes. Focuses on deterministic fixtures, GPU checks, and teardown lifecycle tests.
MIT 681 installs ★ 26 updated 2026-09-10
npx skills add cesartevisual/threejs-skills --skill threejs-testing -a claude-codeWhat threejs-testing does
Guides the agent to test Three.js applications through browser fixtures and production builds instead of relying on mocked renderers. A mocked renderer cannot verify actual GPU pixel output. The skill directs tests to seed random values, lock camera coordinates, viewport size, and pixel ratios, and freeze animations at an explicit timestamp.
The agent waits for semantic asset-loading events rather than arbitrary sleep delays. Image assertions rely on tolerant comparisons because GPU drivers produce minor pixel differences across machines. Verification also covers resize handling, touch inputs, teardown lifecycle leaks, and browser console errors.
What it helps with
- set up reproducible visual regression tests for Three.js scenes
- test scene teardown and resource disposal
- verify rendering after resize and orientation changes
- catch console and network failures during 3D asset loading
Example prompts for the agent
- Write a deterministic Playwright visual snapshot test for this Three.js canvas
- Add lifecycle checks to verify scene teardown and memory cleanup
- Test canvas resize behavior across different device pixel ratios
How to install threejs-testing
1Skills CLI (Vercel)
npx skills add cesartevisual/threejs-skills --skill threejs-testing -a claude-code2Manually: put the files in place
~/.claude/skills/threejs-testing/SKILL.mdCopy the whole skill folder; for one project use .claude/skills/threejs-testing/
1Skills CLI (Vercel)
npx skills add cesartevisual/threejs-skills --skill threejs-testing -a codex2Manually: put the files in place
.agents/skills/threejs-testing/SKILL.mdFor all projects use ~/.agents/skills/
1Skills CLI (Vercel)
npx skills add cesartevisual/threejs-skills --skill threejs-testing -a cursor2Manually: put the files in place
.cursor/skills/threejs-testing/SKILL.mdCursor also reads .agents/skills/ and .claude/skills/
1GitHub CLI
gh skill install cesartevisual/threejs-skills skills/threejs-testing2Manually: put the files in place
.github/skills/threejs-testing/SKILL.md1Skills CLI (Vercel)
npx skills add cesartevisual/threejs-skills --skill threejs-testing -a gemini-cli2Platform CLI
gemini skills install https://github.com/cesartevisual/threejs-skills --path skills/threejs-testingManually: put the files in place
.devin/skills/threejs-testing/SKILL.mdThe legacy .windsurf/skills/ path still works
Skills CLI (Vercel)
npx skills add cesartevisual/threejs-skills --skill threejs-testing -a opencodeClaude.ai
Zip the folder and upload it in Settings → Capabilities → Skills
Step by step, with commands for each platform: How to install skills in Claude Code: five ways
Questions about threejs-testing
How to install threejs-testing in Claude Code?
Run npx skills add cesartevisual/threejs-skills --skill threejs-testing -a claude-code or save the file to ~/.claude/skills/threejs-testing/SKILL.md.
What does threejs-testing do?
It directs coding agents to write reliable Three.js tests using browser fixtures, seeded randomness, tolerant screenshot comparisons, and lifecycle checks.
Why does the skill avoid relying solely on mocked renderers?
A mocked renderer cannot confirm that GPU pixels render accurately. Reliable GPU integration requires running tests against a production build in an actual browser fixture.
Source text
SKILL.md · MIT
--- name: threejs-testing description: "Test Three.js features with browser checks, deterministic visual fixtures, lifecycle tests, and device-specific evidence." --- # 3D Testing and Visual QA Use the project's installed versions and existing architecture. These instructions supplement the user's task; they do not authorize unrelated changes or external actions. Translate the requested behavior into observable checks: visible subject, correct framing, interaction outcomes, load failure/retry, resize, teardown, and any explicit performance target. Keep test scope proportional to the change. Use unit tests for pure coordinate transforms, selection mapping, timestep policies, and reducers where these are meaningful. A mocked renderer cannot prove pixels render correctly. Run the production build and a browser fixture for GPU integration. Make visual fixtures reproducible: seed randomness, fix camera/viewport/pixel ratio, stop or seek animation to a known time, and wait for asset readiness plus a rendered frame. Use a semantic readiness signal, not an arbitrary sleep. Capture console and network failures along with screenshots. Use tolerant comparisons for GPU-dependent pixels; avoid treating exact image equality across unrelated drivers as a portable contract. Inspect screenshots for cropping, contrast, shadow defects, text/overlay occlusion, and loading remnants. Scope browser feature expectations to the tested backend. Exercise repeated navigation, rapid asset changes, hidden-to-visible sizing, touch interaction, reduced motion, unsupported rendering, and the relevant real device/browser. Headless software rendering is not a mobile GPU benchmark. Report checks as passed, failed, or not run with reasons. Do not claim performance, XR hardware support, or assistive-technology behavior from compilation alone. ## API reference Consult the [official documentation](https://playwright.dev/docs/test-snapshots) for exact signatures and revision-specific behavior. If documentation access is unavailable, inspect installed types/source and state any remaining uncertainty.
© cesartevisual · MIT · cesartevisual/threejs-skills