link-crawler
Pass
Audited by Gen Agent Trust Hub on Jun 25, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The agent is instructed to execute the provided
crawl_links.pyscript to perform crawling and link validation tasks. - [EXTERNAL_DOWNLOADS]: The script performs network requests to fetch HTML content from external URLs to verify if links are active and to discover new links.
- [PROMPT_INJECTION]: The skill processes untrusted HTML and URL data from external websites, creating a surface for indirect prompt injection.
- Ingestion points: Untrusted web content is ingested via the
fetchfunction incrawl_links.py. - Boundary markers: The script produces structured JSON output which the agent is instructed to format into a markdown list, providing some structural separation.
- Capability inventory: The skill can perform network operations (
urllib.request) and write output to the local file system (--outputflag). - Sanitization: The script parses HTML to extract links but does not filter for or sanitize potential natural language instructions embedded in the crawled content or URLs.
Audit Metadata