diagnose-hardware-root-causes-rd
Pass
Audited by Gen Agent Trust Hub on Aug 27, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [COMMAND_EXECUTION]: The skill invokes a local Python utility,
scripts/ceae_tree_export.py, to transform JSON-based causal models into PNG visualizations. This execution is intentional, occurs within the skill's own environment, and is restricted to user-defined input and output paths. - [INDIRECT_PROMPT_INJECTION]: The diagnostic workflow processes external hardware data and JSON trees, representing a potential surface for indirect injection.
- Ingestion points: Hardware parameter data provided by the user in
SKILL.mdand structured JSON data loaded viascripts/ceae_tree_export.py. - Boundary markers: The visualization script enforces strict schema validation, including unique node IDs and an allowed list of node types.
- Capability inventory: Performs local file reading/writing and image rendering using standard libraries.
- Sanitization: The script implements
clean_textto remove null bytes and validates the graph structure (acyclic and connected) before processing. - [EXTERNAL_DOWNLOADS]: The skill relies on common third-party Python libraries, specifically
matplotlibandnetworkx. These are widely recognized, well-known scientific packages used for data visualization and graph theory.
Audit Metadata