nvidia-omniverse/usd-content-agents/content-sdf-operations

Claude Code and agent skill by nvidia-omniverse/usd-content-agents

Trust tier 2
  • community
LLM review pending
Source check pending

Checking that the SKILL.md file is still in the repository.

Use when a content agent needs bounded signed- or unsigned-distance-field operations such as mesh rasterization, field booleans, offsets, filtering, sampling, topology conversion, surface extraction, or admitted field output through an in-process backend.

Install

npx vskill@latest install nvidia-omniverse/usd-content-agents/content-sdf-operations

SKILL.md

Content SDF Operations

Use the backend-neutral sdf_tools contract for signed- and unsigned-distance field work. Select semantic capabilities first; treat the implementation that provides them as recorded execution evidence, not as the capability surface.

When To Use

  • Convert a world-space polygon mesh to a signed or unsigned distance field.
  • Apply bounded field booleans, offsets, smoothing, normalization, or rebuilds.
  • Sample field values or gradients at declared world-space points.
  • Convert active topology to an SDF, extract an enclosed region, or polygonize

an isosurface.

  • Write a backend-supported field artifact with an explicit format.
  • Inspect which admitted backend and exact implementation executed an operation.

Use a qualified Geometry Repair workflow instead when the goal is to accept or publish repaired geometry. A successful SDF operation does not establish source fidelity, protected-feature preservation, correspondence, or repair acceptance.

Ownership Boundary

This is an atomic capability, not a workflow. It owns one admitted in-process SDF session and the bounded operations requested from that session. It does not own run state, artifact lifecycle, refinement policy, geometry acceptance, or a terminal workflow outcome. The calling workflow owns those decisions and must bind this capability's execution evidence into its own durable handoff.

Core Contract

python
import numpy as np
import sdf_tools
mesh = sdf_tools.Mesh(
vertices=np.asarray(vertices, dtype=np.float32),
triangles=np.asarray(triangles, dtype=np.int32),
)
session = sdf_tools.create_session(
backend="auto",
require={
sdf_tools.Operation.MESH_TO_SDF,
sdf_tools.Operation.SAMPLE_VALUES,
sdf_tools.Operation.FIELD_TO_MESH,
},
)
field = session.mesh_to_sdf(mesh, voxel_size=0.01, half_width=3.0)
values = session.sample_values(field, sample_points)
surface = session.field_to_mesh(field)
backend_evidence = session.backend_info.as_dict()

backend="auto" is for exploratory work. For a benchmark, qualified geometry route, or durable artifact, select the exact backend admitted by the governing policy and persist backend_info. The first production-qualified driver is openvdb, backed by the source-locked OpenVDB 13 runtime; future admitted drivers may implement the same operations.

Rules

  • Keep world units explicit. Signed fields use negative values inside.
  • Declare voxel size, band width, interpolation, isovalue, thread count, and

resource limits rather than relying on an implementation default.

  • Treat Field as an opaque, backend-owned in-process value. Do not unwrap its

payload, mix fields from different backends, or serialize a live field handle.

  • Request every operation and artifact format needed when creating a session,

using require_formats together with READ_FIELDS or WRITE_FIELDS so the direction is explicit. An unavailable capability must fail admission; it is not permission to substitute another algorithm or lower-level API silently.

  • Do not import openvdb or openvdb_runtime from content-agent code.
  • Do not install, import by module path, or register a model- or caller-supplied

backend. Installed discovery admits only exact entries in checked-in policy.

  • Driver calls execute inside the importing CPython process. Do not add an SDF

helper process, daemon, RPC service, JSON transport, or file-based bridge.

  • Do not admit an SDF dependency delta that introduces or bundles an LGPL

component. Require the structured license manifest and native-closure gate; a top-level package license alone is insufficient.

  • Preserve resource-limit, unavailable, invalid-geometry, and backend-operation

failures as distinct outcomes.

Evidence

For durable outputs, record the requested operations, selected backend ID, backend_info, execution limits, units, voxel and band controls, input/output digests, and any selection rejections. Record the explicit artifact format for field output. Backend identity is provenance, not a replacement for geometric validation.

References

Security

  • Security scanPassed, 100 of 100Pattern checks on the skill's files.
  • Model reviewNot runA model review is scheduled for the next rescan.

Quality evaluation

  • Task evaluationNot runNo evaluation yet. Run one in Skill Studio.

Source

Popularity

Repository stars
205
Repository forks
25
7-day trend score
27

Stars and forks are for the whole GitHub repository, not this skill alone. Not refreshed yet.

Works with

Any agent vskill installs to: 56 targets in all. These 39 have a page on this site.

  • Claude Code
  • Cursor
  • GitHub Copilot
  • Windsurf
  • Codex
  • Gemini CLI
  • Cline
  • Amp
  • Roo Code
  • Goose
  • Continue
  • Kilo Code
  • Junie
  • Trae
  • OpenHands
  • Qwen Code
  • AI Agent
  • Tabnine Chat
  • Sourcegraph Cody
  • Amazon Q
  • Firebase Genkit
  • Sweep
  • Devin
  • Phind
  • Mentat
  • Claude Desktop
  • Aider
  • GPT Pilot
  • Plandex
  • bolt.new
  • v0
  • Replit Agent
  • Anse
  • TypingMind
  • BetterChatGPT
  • LibreChat
  • Warp AI
  • Pieces
  • Fig AI
Embed badge
vskill VERIFIED badge
[![vskill VERIFIED](https://verified-skill.com/api/v1/skills/nvidia-omniverse/usd-content-agents/content-sdf-operations/badge)](https://verified-skill.com/skills/nvidia-omniverse/usd-content-agents/content-sdf-operations)

Something wrong with this skill or its page? Tell us.