bio-data-visualization-network-visualization
Pass
Audited by Gen Agent Trust Hub on Oct 8, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADSCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests external interaction data which could potentially contain malicious instructions. While the visualization tools do not execute these instructions, the presence of the data in the agent's context creates a surface for indirect prompt injection. * Ingestion points: Data is imported from external files (e.g., 'string_interactions.csv') using
pandas.read_csvinexamples/network_plots.pyandexamples/cytoscape_automation.py. * Boundary markers: There are no explicit markers or instructions to ignore embedded content within the processed network data. * Capability inventory: The skill uses tools with file-writing capabilities acrossexamples/network_plots.py,examples/interactive_network.py, andexamples/cytoscape_automation.py. * Sanitization: No sanitization is performed on node labels or metadata attributes before they are rendered or exported. - [EXTERNAL_DOWNLOADS]: The skill references and recommends installation of well-known Python packages (
networkx,pyvis,py4cytoscape,matplotlib,numpy,datashader,pandas) and R packages (igraph,ggraph,tidygraph) for network analysis and visualization. - [COMMAND_EXECUTION]: The documentation suggests using
pip showto check installed package versions. Furthermore, the skill leverages NetworkX's integration with Graphviz (pydot), which involves executing external binaries for graph layout generation.
Audit Metadata