web-searching
Pass
Audited by Gen Agent Trust Hub on Oct 9, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill retrieves and processes untrusted data from external web search providers which is then presented to the agent.
- Ingestion points: Data enters the skill via the
WebSearchService.search()method inhandler.ts, which fetches snippets and titles from third-party search providers. - Boundary markers: The instructions in
SKILL.mddo not define clear delimiters or provide the agent with instructions to ignore potential commands embedded within the search results. - Capability inventory: The skill is limited to search and does not possess dangerous capabilities like file system modification, shell execution, or privilege escalation.
- Sanitization: The results are mapped directly into markdown format in
handler.tswithout sanitization or filtering to detect or neutralize malicious instructions contained within search snippets.
Audit Metadata