The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external data from Yahoo Finance RSS feeds and SEC EDGAR Atom feeds. This creates a potential surface for indirect prompt injection if an attacker-controlled filing or headline contains malicious instructions.
Ingestion points: Data is fetched via urllib.request in news.py from finance.yahoo.com and sec.gov.
Boundary markers: There are no explicit delimiters used in the prompts when presenting news items to the agent.
Capability inventory: The skill is restricted to reading and writing within the trader/ directory and performing network GET requests to specific financial domains. It does not have arbitrary shell execution capabilities.
Sanitization: The script uses xml.etree.ElementTree for structured parsing, which prevents common XML attacks, but textual content (titles/summaries) is passed directly to the agent without further sanitization for instructions.

trader-news