us-stock-analysis

Pass

Audited by Gen Agent Trust Hub on Mar 2, 2026

Risk Level: SAFEPROMPT_INJECTION
Full Analysis
  • [PROMPT_INJECTION]: The skill possesses a vulnerability surface for indirect prompt injection due to its core function.
  • Ingestion points: Untrusted data enters the agent context via web search results for stock metrics and news as defined in the Search Strategy section of SKILL.md.
  • Boundary markers: There are no explicit delimiters or system instructions to treat the search content as untrusted.
  • Capability inventory: The skill is informational and focuses on generating reports; it does not define dangerous capabilities such as file system writing, network exfiltration beyond search, or subprocess execution.
  • Sanitization: There is no evidence of logic for sanitizing or validating external data before it is analyzed by the model.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 2, 2026, 10:32 PM