database-cloud-optimization-cost-optimize

Pass

Audited by Gen Agent Trust Hub on Feb 17, 2026

Risk Level: SAFEPROMPT_INJECTION
Full Analysis
  • Indirect Prompt Injection (LOW): The skill instructs the agent to analyze external cloud spending and resource data, creating a potential vector for malicious instructions embedded in resource names or billing metadata.\n
  • Ingestion points: Collection of cost data by service and resource as specified in SKILL.md.\n
  • Boundary markers: Absent; there are no instructions to use delimiters or ignore embedded commands in the ingested data.\n
  • Capability inventory: The skill allows for implementing budgets, alerts, and performing resource resizing or deletion as per SKILL.md.\n
  • Sanitization: Absent; the skill does not define methods for escaping or validating external data before processing.
Audit Metadata
Risk Level
SAFE
Analyzed
Feb 17, 2026, 06:07 PM