pkpd-modeling

Pass

Audited by Gen Agent Trust Hub on Sep 3, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection due to its processing of untrusted external data files and its access to high-capability tools.
  • Ingestion points: The skill ingests data from external CSV and TSV files via the read_table function in scripts/_common.py. This data is used as input for all primary modeling scripts.
  • Boundary markers: The data processing pipeline lacks explicit delimiters or instructions to the agent to ignore embedded natural language commands within the data fields.
  • Capability inventory: The skill is configured with broad tool access in SKILL.md (allowed-tools: Read Write Edit Bash) and includes scripts that utilize numpy and scipy for complex calculations.
  • Sanitization: The _common.parse_float function provides a basic level of sanitization by enforcing that processed data must be numeric; however, this does not prevent the agent from being influenced by the raw content of files if read directly using other tools.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 3, 2026, 04:56 AM
Security Audit — agent-trust-hub — pkpd-modeling