signals-scout-health-checks
Pass
Audited by Gen Agent Trust Hub on Oct 23, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill identifies a potential attack surface by processing untrusted data from health check payloads and ingestion logs.
- Ingestion points: Data enters the context through
health-issues-get(payload, title, summary) andexecute-sqlqueries against thesystem.ingestion_warningstable. - Boundary markers: The skill includes a robust 'Untrusted data — payload fields' section that explicitly warns the agent to ignore any instructions embedded in data and treats remediation fields as the only trusted guidance.
- Capability inventory: The skill has access to write tools including
scout-emit-report,scout-edit-report, andscout-scratchpad-remember. - Sanitization: Instructions require quoting untrusted snippets, using stable identifiers (UUIDs) for keys, and avoiding the use of free-text fields for decision-making logic.
- [COMMAND_EXECUTION]: The skill utilizes SQL queries via
execute-sqlto analyze system telemetry and traffic patterns. This capability is used to validate the impact (blast radius) of detected issues, and the queries are structured as predefined templates for analytics rather than arbitrary execution.
Audit Metadata