landing-page-intel
Pass
Audited by Gen Agent Trust Hub on Oct 10, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests and processes HTML content from external, untrusted websites, creating a potential surface for indirect prompt injection where an attacker could embed instructions in the target website's HTML.
- Ingestion points: The
scripts/scrape_landing_page.pyscript fetches external content using therequests.getmethod. - Boundary markers: The skill does not explicitly define prompt delimiters or provide instructions to the agent to ignore embedded commands within the scraped HTML.
- Capability inventory: The script's capabilities are limited to network reads (
requests.get) and outputting parsed data to stdout. It does not include file system writes, privilege escalation, or dynamic code execution. - Sanitization: The script uses regular expressions to strip HTML tags from extracted text, providing a basic layer of sanitization for display.
- [COMMAND_EXECUTION]: The skill requires the agent to execute a local Python script (
scripts/scrape_landing_page.py) to perform its primary function. This is an intended behavior for a capability-based skill and involves standard command-line argument passing.
Audit Metadata