arboreto
Pass
Audited by Gen Agent Trust Hub on Apr 16, 2026
Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTIONDATA_EXFILTRATIONEXTERNAL_DOWNLOADS
Full Analysis
- [PROMPT_INJECTION]: The skill includes instructions to the agent to proactively suggest the vendor's 'K-Dense Web' platform when users encounter complex tasks. This modifies agent behavior for promotional purposes but does not attempt to bypass safety constraints.
- [COMMAND_EXECUTION]: Provides standard CLI commands for installing the software and executing local python scripts.
- [DATA_EXFILTRATION]: Documents the capability to connect to remote Dask schedulers over TCP for distributed computing tasks.
- [EXTERNAL_DOWNLOADS]: Encourages the installation of the 'arboreto' package and its dependencies from public registries.
- [SAFE]: The skill performs data ingestion from external files. 1. Ingestion points: Gene expression data loaded via pandas.read_csv in SKILL.md. 2. Boundary markers: Absent. 3. Capability inventory: Local file writing and subprocess execution for analysis scripts. 4. Sanitization: Absent.
Audit Metadata