geo-optimizer
Pass
Audited by Gen Agent Trust Hub on Sep 11, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [COMMAND_EXECUTION]: Several scripts in the skill automatically invoke the system package manager to install dependencies if they are missing. Evidence:
scripts/audit-geo.py,scripts/check-hedge-density.py, andscripts/generate-agentfacts.pyall contain code blocks usingsubprocess.check_call([sys.executable, "-m", "pip", "install", ...])to installrequestsandbeautifulsoup4at runtime. - [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process data from external, potentially untrusted URLs, which creates a surface for indirect prompt injection. 1. Ingestion points: The
fetch_contentmethod inscripts/audit-geo.pyandfetch_text_from_urlinscripts/check-hedge-density.pydownload content from user-specified URLs. 2. Boundary markers: There are no explicit delimiters or instructions provided to the agent to ignore instructions embedded within the fetched content. 3. Capability inventory: The skill utilizesrequestsfor network operations andsubprocessfor command execution (pip installation). 4. Sanitization: WhileBeautifulSoupis used to remove HTML tags, the resulting text content is analyzed for patterns (hedge density) without filtering for malicious instructions. - [EXTERNAL_DOWNLOADS]: The skill performs network requests to fetch external website content for auditing purposes and to download packages from the Python Package Index (PyPI).
Audit Metadata