signals-scout-data-pipelines
Pass
Audited by Gen Agent Trust Hub on Oct 6, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes data from function logs, error messages, and recipient outcomes that can contain untrusted content derived from event payloads or remote server responses. This constitutes a potential indirect prompt injection surface. The skill includes specific defensive instructions to mitigate this risk.
- Ingestion points: Data enters the context through
cdp-functions-logs-retrieve,workflows-logs, andworkflows-list-invocations. - Boundary markers: Instructions explicitly direct the agent to treat this data as untrusted and pair it with counts for verification.
- Capability inventory: The skill uses
execute-sql,scout-emit-report,scout-edit-report, andscout-scratchpad-remember. - Sanitization: The agent is instructed to truncate long messages and drop payload echoes when citing errors in reports.
Audit Metadata