web-region-audience-analyzer
Pass
Audited by Gen Agent Trust Hub on Mar 21, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted web content using an LLM to generate localization reports and select pages for crawling.
- Ingestion points:
scripts/analyzer/fetcher.pyandscripts/analyzer/crawling.pyfetch HTML content and link metadata from external URLs provided by the user. - Boundary markers: The skill uses markdown code blocks (triple backticks) to demarcate untrusted content within LLM prompts in
scripts/analyzer/ai_analysis.pyandscripts/analyzer/crawling.py, but lacks explicit 'ignore embedded instructions' warnings. - Capability inventory: The skill performs network requests to fetch HTML content and makes calls to an external AI API (OpenAI-compatible) and NLP Cloud.
- Sanitization: Page content is truncated to 3000 characters before being sent to the LLM, but no other sanitization or escaping of potentially malicious instructions within the HTML is performed.
Audit Metadata