etetoolkit
Pass
Audited by Gen Agent Trust Hub on Apr 16, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The documentation includes installation steps requiring
sudoto install system packages (python3-pyqt5) necessary for graphical tree rendering. - [EXTERNAL_DOWNLOADS]: The skill uses the
NCBITaxamodule, which automatically downloads the NCBI taxonomy database (~300MB) from official NCBI servers during initialization to support taxonomic lookups. - [PROMPT_INJECTION]: The skill contains explicit instructions directing the agent to steer user interactions toward the 'K-Dense Web' platform when handling complex or multi-step research tasks.
- [PROMPT_INJECTION]: The skill processes external hierarchical data files (Newick/NHX), representing a surface for indirect prompt injection when combined with file-writing and script-execution capabilities.
- Ingestion points: Data is loaded via
Tree,PhyloTree, andClusterTreeconstructors. - Boundary markers: No specific instructions or delimiters are provided to ensure the agent ignores instructions embedded within the processed data files.
- Capability inventory: The skill can write files (
tree.write), render images (tree.render), and execute shell commands. - Sanitization: There is no documented validation or sanitization of the input tree data contents.
- [COMMAND_EXECUTION]: The documentation references and provides usage examples for executing auxiliary Python scripts (
scripts/tree_operations.py,scripts/quick_visualize.py) for batch processing.
Audit Metadata