news-sentiment-engine

Fail

Audited by Gen Agent Trust Hub on Aug 24, 2026

Risk Level: HIGHEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill provides setup instructions to download code from an unverified GitHub repository (https://github.com/tellmefrankie/news-engine).- [COMMAND_EXECUTION]: The skill includes shell commands (git clone, pnpm install, pnpm dev) that execute the downloaded external code locally on the user's system.- [PROMPT_INJECTION]: The skill ingests untrusted data from multiple RSS feeds, creating a surface for indirect prompt injection attacks.
  • Ingestion points: Aggregates news content from various external RSS feeds (TechCrunch, The Verge, Ars Technica, Hacker News).
  • Boundary markers: Missing clear delimiters or instructions to prevent the agent from following directions embedded within the fetched articles.
  • Capability inventory: The skill uses claude and websearch tools and requests local shell execution via the setup commands.
  • Sanitization: There is no mention of filtering, escaping, or validating the external content before it is processed by the agent.
Recommendations
  • AI detected serious security threats
Audit Metadata
Risk Level
HIGH
Analyzed
Aug 24, 2026, 05:59 AM
Security Audit — agent-trust-hub — news-sentiment-engine