marketing-guerrilla

Warn

Audited by Gen Agent Trust Hub on Mar 13, 2026

Risk Level: MEDIUMEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill instructs the agent to install the agent-browser package and Microsoft Playwright browser binaries from the NPM registry if they are not already available in the environment.
  • [COMMAND_EXECUTION]: Shell commands are used to perform global software installations (npm install -g) and binary setup (npx playwright install), which may require elevated system privileges.
  • [PROMPT_INJECTION]: The skill possesses a high surface for indirect prompt injection because it scrapes and extracts information from various external, user-controlled platforms including Reddit, TikTok, YouTube, and Product Hunt.
  • Ingestion points: Web content is fetched from Google Trends, YouTube, Reddit, TikTok, and Product Hunt URLs using the agent-browser tool in SKILL.md.
  • Boundary markers: There are no boundary markers or instructions to treat external data as untrusted text rather than agent instructions.
  • Capability inventory: The agent has the capability to write to the local file system (./brands/) and execute package management commands.
  • Sanitization: No logic is present to sanitize, validate, or filter the scraped content before the agent interprets it for viral trends and strategy insights.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Mar 13, 2026, 04:01 AM