signals-scout-health-checks
Pass
Audited by Gen Agent Trust Hub on Aug 4, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: Indirect Prompt Injection Surface. The skill is designed to process "health issues" which include data fields like payload, title, and summary that are populated by external project or event data. This creates a surface where an attacker could potentially embed malicious instructions in the health issue data to influence the agent's behavior.
- Ingestion points: Data is ingested via the health-issues-get and ingestion-warnings-list tools, which return fields like payload, title, summary, and details populated by external entities.
- Boundary markers: The skill documentation includes a dedicated "Untrusted data — payload fields" section that explicitly warns the agent to treat these values as data and not instructions.
- Capability inventory: The agent has access to tools for reporting (scout-emit-report, scout-edit-report) and data querying (execute-sql, query-trends).
- Sanitization: The instructions recommend quoting untrusted snippets and emphasize that external data must not be used to authorize actions or influence decision-making logic.
Audit Metadata