news
Pass
Audited by Gen Agent Trust Hub on Mar 24, 2026
Risk Level: SAFE
Full Analysis
- [COMMAND_EXECUTION]: The skill instructs the agent to utilize existing tools like
search-and-fetch,defuddle,crwl, andagent-browserfor retrieving information. These operations are core to the skill's stated purpose of news aggregation. - [EXTERNAL_DOWNLOADS]: The skill fetches data from external domains including GitHub (a well-known service) and established news organizations such as TechCrunch, Reuters, Bloomberg, and The Verge. These network operations are intended for data retrieval (news content and RSS feeds) rather than executable code.
- [SAFE]: The skill processes untrusted data from the internet (news articles), which is an inherent surface for indirect prompt injection. However, the use of strict summarization rules (25-50 words) and a predefined output template in
references/output-template.mdserves as a natural mitigation against malicious instructions embedded in news content.
Audit Metadata