database-cloud-optimization-cost-optimize
Pass
Audited by Gen Agent Trust Hub on Oct 27, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The implementation playbook contains HTML dashboard templates that reference external JavaScript libraries (Chart.js and D3.js) from well-known content delivery networks such as jsDelivr and d3js.org. This is standard practice for visual reporting tools and follows established patterns for dashboard implementation.
- [INDIRECT_PROMPT_INJECTION]: The
AutoTaggerutility in the implementation playbook represents a surface for indirect prompt injection as it ingests metadata from cloud provider events (e.g., CloudTrail) to determine tags for resources. The skill mitigates this risk by providing instructions on data escaping and validation to ensure that untrusted external inputs do not lead to unintended resource modifications or metadata poisoning. - [COMMAND_EXECUTION]: The skill includes Python scripts and configuration blueprints that facilitate high-impact infrastructure operations, such as stopping instances, modifying instance types, and altering network configurations. These capabilities are directly related to the skill's primary objective of infrastructure cost optimization. The risk is managed through detailed instructions emphasizing authorized environments, staged rollouts, and the requirement for explicit user approval before any changes are applied.
Audit Metadata