web-search
Pass
Audited by Gen Agent Trust Hub on Sep 17, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted content from the internet through search results, which is a common attack surface for indirect prompt injection.
- Ingestion points: The
scripts/search.pyscript ingests data (titles, descriptions, URLs) directly from external websites via the DuckDuckGo search API. - Boundary markers: The script formats results using standard markers (numbered lists in text/markdown, structured fields in JSON), but it does not include explicit instructions to the AI agent to ignore potentially malicious commands embedded within the search snippets.
- Capability inventory: The skill has the capability to write output to the local file system using the
--outputargument inscripts/search.py. - Sanitization: There is no sanitization or filtering of the retrieved search content to remove or escape instructions that might attempt to manipulate the agent.
Audit Metadata