market-analysis

Pass

Audited by Gen Agent Trust Hub on Feb 28, 2026

Risk Level: SAFEPROMPT_INJECTION
Full Analysis
  • [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection through its data ingestion workflows.
  • Ingestion points: The skill fetches untrusted external data via get_news(ticker), get_global_news(), and Reddit sentiment analysis (referenced in reddit_utils.py).
  • Boundary markers: The prompt lacks explicit instructions for the AI to ignore instructions embedded within the fetched news or social media text.
  • Capability inventory: The skill possesses the capability to generate financial advice and specific trading recommendations (BUY/HOLD/SELL) based on this external data.
  • Sanitization: There is no evidence of sanitization or filtering of the external content before it is processed by the AI models.
Audit Metadata
Risk Level
SAFE
Analyzed
Feb 28, 2026, 04:42 PM