database-cloud-optimization-cost-optimize
Pass
Audited by Gen Agent Trust Hub on Apr 21, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The skill demonstrates a surface for indirect prompt injection by processing external resource metadata that could be controlled by an attacker and using it to perform state-changing operations.
- Ingestion points: The skill is designed to collect and analyze cost data, service identifiers, and resource metadata from cloud environments (SKILL.md).
- Boundary markers: No explicit markers, delimiters, or instructions to ignore potential commands within the ingested metadata are present in the prompt.
- Capability inventory: The skill instructions include the capability to implement budgets and alerts in the target environment (SKILL.md).
- Sanitization: There is no mention of sanitizing or validating the ingested resource metadata before processing.
Audit Metadata