t3-hardware-scoring-cn
Pass
Audited by Gen Agent Trust Hub on Feb 24, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The script
scripts/crawl_product_info.pyfetches web content by making network requests tor.jina.ai. Jina AI is a well-known technology service used to convert web pages into structured text for AI processing. - [COMMAND_EXECUTION]: The
SKILL.mdinstructions require the agent to execute the local Python scriptscripts/crawl_product_info.pyto scrape and parse product specifications and market data. - [PROMPT_INJECTION]: The skill processes external data from arbitrary product URLs, creating an indirect prompt injection surface.
- Ingestion points: Untrusted data is ingested from external URLs via the
crawl_webpagefunction inscripts/crawl_product_info.py. - Boundary markers: The process lacks explicit delimiters or instructions to ignore embedded commands within the scraped text.
- Capability inventory: The agent uses this data to perform scoring, calculate investment advice, and generate final audit reports (
99-audit-report.md). - Sanitization: The 'Brand Blinding' process (detailed in
references/defluff-guide.md) acts as a functional filter by removing marketing fluff and brands, but it is not designed to sanitize against adversarial instructions.
Audit Metadata