signals-scout-web-analytics
Pass
Audited by Gen Agent Trust Hub on Sep 20, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data originating from external web browsers, specifically URL paths, referrer domains, and UTM parameters stored in the
sessionsandeventstables. These fields could be used for indirect prompt injection attacks. - Ingestion points: The skill reads external data via the
execute-sqltool from thesessionsandeventstables. - Boundary markers: The instructions explicitly direct the agent to treat these values as data snippets rather than instructions, using quoting to delimit them.
- Capability inventory: The skill is limited to read-only analytics access and specific reporting tools (
emit_report,edit_report). It does not have access to arbitrary shell commands or external network requests. - Sanitization: The skill mandates the use of sanitized identifiers (slugs) and aggressive truncation of user-supplied strings before they are used in memory or reports.
Audit Metadata