signals-scout-insight-alerts
Pass
Audited by Gen Agent Trust Hub on Sep 10, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests and processes user-defined data such as alert names, insight titles, and report content which could contain malicious instructions.
- Ingestion points: Potential untrusted data enters the agent context via the
alerts-list,alert-get, andinsight-gettools which retrieve project-specific configurations. - Boundary markers: The skill instructions do not specify the use of delimiters or markers to separate retrieved external data from the agent's internal reasoning or report templates.
- Capability inventory: The agent has write capabilities to the project's reporting channel via the
scout-emit-reportandscout-edit-reporttools. - Sanitization: There is no evidence of sanitization or escaping of the ingested strings (e.g., alert names) before they are interpolated into report summaries or evidence fields.
Audit Metadata