nvidia-omniverse/usd-content-agents/content-articulation-inspection
Claude Code and agent skill by nvidia-omniverse/usd-content-agents
- community
Checking that the SKILL.md file is still in the repository.
Inspect bounded USD hierarchy, existing articulation, and usd-cli evidence before proposing joints. Use for exact Articulation candidate preparation and targeted reinspection.
Install
npx vskill@latest install nvidia-omniverse/usd-content-agents/content-articulation-inspectionSKILL.md
Articulation Inspection
Inspect geometry, hierarchy, existing joints, properties, picks, and focused renders without authoring or changing the source asset.
When to Use
Use before candidate proposal and after validation identifies a specific joint or moving-part scope that needs more evidence.
Limitations
- Never mutate the source or infer joints from labels alone.
- Inspect only the request-bound asset and candidate moving-part scope.
- A preview render is diagnostic; saved-stage readback is final evidence.
Prerequisites
Load usd-cli and the frozen Articulation request. Preserve exact retained artifact bindings throughout the run. The coordinator does not derive the package-aware dependency-bundle identity.
Instructions
- Inspect hierarchy, xforms, bounds, mesh topology, and existing joint graph.
- Capture request-bound properties, picks, cameras, and focused renders.
- Verify fixed-parent and moving-part paths are distinct and present.
- Record exact evidence paths and digests for proposal/review.
- For provider-neutral preparation, persist one complete typed
ArticulationPreparationInspectionReadbackDraft with schema content-agent-workflows.articulation-preparation-readback-draft.v2: exact retained source and dependency claims, inspector configuration and implementation, saved stage, lexical hierarchy parents, per-prim membership/owner observations, renders, and scene records. Omit source_dependency_bundle_sha256 and saved_stage_dependency_bundle_sha256; do not calculate, guess, copy, or recover either value from an error. The trusted publisher independently derives and seals both package-aware identities. The v1 readback is only a compatibility input for a trusted producer that already owns both exact identities. The retained inspector configuration is the exact membership authority; its rows must cover the saved hierarchy and equal the readback observations, and every referenced authoritative owner must be self-owned. Set proposal_status to not_evaluated only when the outer policy selected a proposal provider; otherwise retain not_requested. Do not hand-author separate caller member or owner lists. When this preparation feeds the public standalone/asset Articulation author leaf, the retained inspector configuration must set capabilities.canonical_output_evidence_required to true. This declares the required post-author evidence policy; it does not supply a visual envelope, render the pre-author stage, or change the author-before-canonical- render dependency. Save the inspected stage directly into the retained leaf directory with usd-cli-tel --json save <retained-path>.usda --flatten, verify that exact regular file and digest, and bind it as saved_stage. Never bind a path returned by checkpoint save: parent-owned .usd-cli checkpoint state is session-ephemeral and may be removed before the create-only preparation publisher captures it.
- Stop on truncated, missing, or cross-run evidence.
Command Reference
content-workflow-cli articulation _agent-prepare --run-dir <run> performs deterministic inspection, inference, and evidence collection, then pauses.
Common Workflows
On targeted reinspection, retain accepted candidate decisions and collect new evidence only for the failed candidate scope.
Output Format
Use the request, candidate document, usd-cli manifest, exact candidate IDs, and evidence digest map named by articulation_agent_observation.json. The provider-neutral leaf additionally consumes the typed saved readback described in agentic/docs/articulation_preparation_attempt_contract.md.
Troubleshooting
If a prim path or evidence digest does not match, stop before proposal review; do not repair identity by editing an artifact in place.
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
- SKILL.md in the repositoryCheckingagentic/.agents/skills/content-articulation-inspection/SKILL.md
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
[](https://verified-skill.com/skills/nvidia-omniverse/usd-content-agents/content-articulation-inspection)