Back to skill
Skillv0.1.0
ClawScan security
Cad Mesh 3dgs · ClawHub's context-aware review of the artifact, metadata, and declared behavior.
Scanner verdict
BenignApr 30, 2026, 10:46 AM
- Verdict
- benign
- Confidence
- high
- Model
- gpt-5-mini
- Summary
- The skill is internally consistent with its stated purpose (advice and code snippets for CAD/mesh ↔ 3D Gaussian Splatting workflows) and does not request extra credentials, installs, or system access.
- Guidance
- This skill is an instruction-only research/helper bundle and appears coherent. Before using it, remember: (1) the examples reference Python/PyTorch—if you plan to execute them, do so in a controlled environment and install libraries from official sources; (2) the agent may ask you to upload CAD/mesh files to provide concrete help—avoid sharing proprietary or sensitive models unless you trust the destination; (3) the skill contains code snippets for illustration only and will not run anything by itself. If you allow the agent to run code on your machine or cloud, review those execution steps and required dependencies first.
Review Dimensions
- Purpose & Capability
- okName, description, and SKILL.md content align: the skill provides guidance, analysis, and code examples for mesh/CAD ↔ 3DGS workflows. It does not request unrelated binaries, credentials, or config paths.
- Instruction Scope
- okSKILL.md contains explanatory text and illustrative code (Python/PyTorch-style snippets) and stays within the problem domain (conversion methods, initialization strategies, known issues). It does not instruct the agent to read system files, environment variables, or to transmit data to external endpoints. No open-ended instructions request arbitrary context collection.
- Install Mechanism
- okThere is no install spec and no code files—this is an instruction-only skill. That minimizes on-disk execution risk; nothing is downloaded or installed by the skill itself.
- Credentials
- okThe skill declares no required environment variables, credentials, or config paths. The content references libraries (e.g., torch) in examples but does not demand any secrets or unrelated service access.
- Persistence & Privilege
- okFlags are default: always=false and agent invocation is allowed (normal). The skill does not request persistent system presence or modify other skills' configs.