google-image-crawler
Fail
Audited by Gen Agent Trust Hub on Sep 21, 2026
Risk Level: CRITICALINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted metadata from the web that could be used for injection attacks. 1. Ingestion points: The search method in core/crawler.py extracts image titles and source URLs from Google Search results. 2. Boundary markers: No delimiters or warnings are used to separate crawled content from instructions. 3. Capability inventory: The skill has file system write capabilities (core/downloader.py and core/crawler.py) and network request capabilities. 4. Sanitization: Extracted metadata is not sanitized or escaped before being returned to the agent.
- [EXTERNAL_DOWNLOADS]: A URL in the documentation has been flagged as malicious by automated scanners. Evidence: The file references/examples.md contains the URL http://another-site.com/pic.webp, which is blacklisted. While included as part of a validation example, the presence of known malicious domains in skill documentation is a security concern.
Recommendations
- Contains 1 malicious URL(s) - DO NOT USE
Audit Metadata