rf-model-importance-analysis
Pass
Audited by Gen Agent Trust Hub on Sep 23, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes user-supplied CSV and TSV files, creating a surface where untrusted data is ingested into the agent context.
- Ingestion points:
scripts/io.Rreads data from files provided via--input_fileand--group_filearguments. - Boundary markers: The skill does not utilize specific delimiters or instructions to ignore potential commands embedded within the data files.
- Capability inventory: The skill has file-writing capabilities (CSV, PDF, RDS) and can write session metadata (
session_info.txt). All output paths are restricted to the skill root via validation inscripts/path_utils.R. It lacks network access and shell execution capabilities. - Sanitization:
scripts/io.Renforces strict numeric conversion for all feature columns and validates group labels, which prevents malicious text strings from being processed as executable logic by the modeling functions.
Audit Metadata