predictive-modeling

Pass

Audited by Gen Agent Trust Hub on Sep 15, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted user data, creating a potential vector for indirect prompt injection.
  • Ingestion points: scripts/run_fold_table.py and scripts/leakage_smoke.py load data via the --input parameter.
  • Boundary markers: Absent; the skill does not use protective delimiters when the agent interprets the script outputs.
  • Capability inventory: scripts/run_fold_table.py can write files to the local system (--out, --predictions-out) and the skill environment allows shell command execution.
  • Sanitization: Absent; the scripts validate numerical consistency and row integrity but do not sanitize for embedded instructional text in input files.
  • [COMMAND_EXECUTION]: The skill instructions require the agent to perform package installations using pip and execute the provided Python scripts in the shell environment.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 15, 2026, 03:57 PM
Security Audit — agent-trust-hub — predictive-modeling