signals-scout-surveys
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 user feedback from PostHog surveys, creating an attack surface for indirect prompt injection.
- Ingestion points: Survey responses are ingested through
execute-sqlqueries defined inreferences/response-querying.mdand processed in the 'Theme aggregation' section ofSKILL.md. - Boundary markers: The skill lacks explicit instructions to treat response data with boundary markers or specific 'ignore' directives, though it is intended for summarization.
- Capability inventory: The agent's capabilities are restricted to reporting (
emit_report,edit_report) and scratchpad updates, limiting the impact of successful injection to the reporting channel. - Sanitization: The skill implements basic sanitization by instructing the agent to filter out test/placeholder responses and remove Personally Identifiable Information (PII) before inclusion in reports.
Audit Metadata