torch-geometric
Pass
Audited by Gen Agent Trust Hub on Sep 24, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: Fetches library components and pre-compiled wheels from data.pyg.org, the official hosting service for PyTorch Geometric dependencies.
- [EXTERNAL_DOWNLOADS]: Downloads standard machine learning benchmark datasets (e.g., Cora, PubMed, QM9, Reddit) from official research hosting providers when using built-in dataset classes in SKILL.md and scripts/benchmark_model.py.
- [COMMAND_EXECUTION]: Provides instructions for installing the environment and executing utility scripts for visualization and template generation.
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted graph data from CSV files and external datasets, which is a standard surface for machine learning tools.
- Ingestion points: pd.read_csv in SKILL.md and built-in dataset loading functions across the skill body and scripts.
- Boundary markers: None present in the data ingestion examples.
- Capability inventory: File system write operations in scripts/create_gnn_template.py and visualization output in scripts/visualize_graph.py.
- Sanitization: No explicit sanitization or validation of graph attributes is performed during the ingestion phase.
Audit Metadata