pp-drug-enforcement

Pass

Audited by Gen Agent Trust Hub on Jul 28, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONDATA_EXFILTRATIONPROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill requires the installation of a CLI tool from vendor-owned sources. It provides instructions to use npx -y @mvanhorn/printing-press-library and go install github.com/mvanhorn/printing-press-library/library/health/drug-enforcement/cmd/drug-enforcement-pp-cli@latest.
  • [COMMAND_EXECUTION]: The skill operates by executing shell commands via the drug-enforcement-pp-cli binary. It includes a which command that uses natural language to resolve command mappings, adding a layer of dynamic command selection.
  • [DATA_EXFILTRATION]: The skill exposes a data exfiltration surface through the --deliver webhook:<url> flag, which allows command output to be sent to arbitrary external URLs. Furthermore, the feedback command can be configured to POST local data to a remote endpoint if the DRUG_ENFORCEMENT_FEEDBACK_ENDPOINT environment variable is set.
  • [PROMPT_INJECTION]: An indirect prompt injection surface exists because the skill ingests and processes data from the openFDA API. Maliciously crafted content within the FDA's recall records could potentially influence the agent's logic when the data is parsed or summarized.
  • Ingestion points: External openFDA API (/drug/enforcement.json).
  • Boundary markers: None explicitly defined in the provided instructions for the processed data.
  • Capability inventory: Subprocess execution via Bash for CLI tool operations.
  • Sanitization: The skill instructions specify that the agent should only report facts and never declare a drug safe, but no technical sanitization of the API string content is described.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 28, 2026, 09:18 AM
Security Audit — agent-trust-hub — pp-drug-enforcement