storm-research

Pass

Audited by Gen Agent Trust Hub on Aug 17, 2026

Risk Level: SAFEPROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill creates a surface for indirect prompt injection by interpolating untrusted user data into analytical prompts and file paths.
  • Ingestion points: The user-provided 'TOPIC' input is utilized throughout the four research phases in SKILL.md and in the final output file path.
  • Boundary markers: No delimiters or instructions are provided to the model to ignore embedded commands within the user-provided research topic.
  • Capability inventory: The skill has the capability to write research artifacts to the local filesystem at 'Research/STORM/'.
  • Sanitization: No sanitization or path-validation logic is applied to the user input before it is used in prompts or file operations.
  • [PROMPT_INJECTION]: Malicious instructions embedded in the research topic could attempt to divert the agent's behavior or bias the synthesis of perspectives, though the structured four-phase process and self-critique phase provide some inherent mitigation.
  • [UNVERIFIABLE_DEPENDENCIES_AND_REMOTE_CODE_EXECUTION]: The skill refers to resources from Stanford University's OVAL research group (GitHub and official website). These are well-known academic sources and the references are documented neutrally as they do not involve untrusted remote code execution.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 17, 2026, 11:52 AM
Security Audit — agent-trust-hub — storm-research