agentleFS
mirzaaghazadehmirzaaghazadeh/iwsdk-webxr-skills/skills/iwsdk-official/iwsdk-depth-occlusion/SKILL.md

iwsdk-depth-occlusion

Skill1 starsChanged 32 days agoMITView on GitHub

Add or repair IWSDK WebXR depth sensing and real-world occlusion in AR. Use for DepthSensingSystem, DepthOccludable, occlusion modes, or passthrough depth troubleshooting.

The file

We split this skill into the rules it gives an agent

9 rules. About 0 are common, 4 are distinctive, and 5 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈4

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 5

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    Keep world.xr.mode as "ar".

    line 20
  2. Very specific:

    Import and register DepthSensingSystem exactly once after World.create:

    line 32
  3. Very specific:

    Add DepthOccludable only to the requested scene nodes.

    line 46
  4. Distinctive:

    Leave a deliberate non-occludable reference object when comparison is part of the request.

    line 52
  5. Distinctive:

    Run the production build once, start the AR app once, enter XR, and query the named entities and registered systems.

    line 57
  6. Distinctive:

    Use one XR entry, one bounded entity/system query set, and one console check.

    line 65
  7. Very specific:

    Use the CLI action names below.

    line 70
  8. Very specific:

    Use npx @iwsdk/cli browser screenshot --output-file <path> for file evidence.

    line 83
  9. Distinctive:

    Do not inspect framework source when the build, registered-system query, entity components, and console warnings already establish the …

    line 86

≈ “Common” is an estimate: at least 1 in 5 of the 372,429 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

IWSDK Depth Occlusion

Depth occlusion has three independent requirements: the AR session requests depth, one depth system is registered, and only intended virtual entities carry DepthOccludable.

Implement

  1. Inspect iwsdk.config.json, src/index.ts, and the active scene in one bounded pass, then make the complete config/system/scene edit before starting the dev server. The contracts below are sufficient for the normal path: do not call scene capabilities, CLI --help, inspect node_modules, or read generated declarations unless a concrete build/runtime error remains.
  2. Keep world.xr.mode as "ar". Under world.xr.features, request:

    "depthSensing": {
      "required": true,
      "usage": "gpu-optimized",
      "format": "float32"
    }

    Preserve other required AR features. Use the living-room emulator (or the environment requested by the task) so runtime depth exists.

  3. Import and register DepthSensingSystem exactly once after World.create:

    world.registerSystem(DepthSensingSystem, {
      configData: {
        enableDepthTexture: true,
        enableOcclusion: true,
        useFloat32: true,
        blurRadius: 20,
      },
    });

    DepthOccludable is built in. Do not call registerComponent for it.

  4. Add DepthOccludable only to the requested scene nodes. These scene payloads are exact: {} means soft occlusion and { "mode": "HardOcclusion" } means sharp low-cost occlusion. Use "MinMaxSoftOcclusion" only when the task needs the higher-quality path. Write the literal scene value directly; do not load the scene-composer format reference, import OcclusionShadersMode, or inspect its declaration to reconfirm it.
  5. Leave a deliberate non-occludable reference object when comparison is part of the request. Do not mark UI or guidance surfaces occludable by default.

Verify

Run the production build once, start the AR app once, enter XR, and query the named entities and registered systems. Check the console for missing depth-sensing warnings. Capture one runtime screenshot in the configured room; source inspection or an authored scene render alone cannot prove passthrough occlusion. Make one focused correction if the measured runtime disagrees, then stop. Do not repeatedly restart the emulator or tune blur from screenshots when the session never exposed depth data.

Use one XR entry, one bounded entity/system query set, and one console check. Save the final live screenshot but do not read it back when those runtime checks already prove that depth was requested, the system is live, and the intended entities carry the correct occlusion modes.

Use the CLI action names below; do not substitute MCP method names or probe --help first:

npx @iwsdk/cli xr enter
npx @iwsdk/cli xr status
npx @iwsdk/cli ecs find --input-json '{"namePattern":"<requested node regex>"}'
npx @iwsdk/cli ecs query --input-json '{"entityIndex":<index from find>}'
npx @iwsdk/cli ecs systems
npx @iwsdk/cli browser logs --count 80
npx @iwsdk/cli browser screenshot --output-file artifacts/<name>.png

Use npx @iwsdk/cli browser screenshot --output-file <path> for file evidence. Do not alter the project camera or render extra authored angles to manufacture an occlusion view; stage the requested nodes once, then judge the live AR capture. Do not inspect framework source when the build, registered-system query, entity components, and console warnings already establish the failing layer.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.