monte-carlo-monitor-creation
Pass
Audited by Gen Agent Trust Hub on Oct 26, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data sourced from external databases, creating a potential surface for indirect prompt injection.
- Ingestion points: External metadata including table names, column names, and schema details are retrieved using the
getTableandsearchtools as described inSKILL.md(Steps 2 and 3) and various reference files. - Boundary markers: The instructions lack explicit delimiters or "ignore embedded instructions" guards when interpolating retrieved metadata into the agent's reasoning or the final generated YAML/SQL.
- Capability inventory: The skill uses MCP tools to generate complex monitoring configurations and SQL queries (
createCustomSqlMonitorMac,createMetricMonitorMac, etc.). - Sanitization: While the instructions emphasize using grounded data from the tools and require user confirmation before calling creation tools, there is no technical sanitization or escaping of the ingested strings before they are incorporated into the output.
Audit Metadata