AWS Cost Optimizer-hardened
Pass
Audited by Gen Agent Trust Hub on Apr 21, 2026
Risk Level: SAFE
Full Analysis
- [PROMPT_INJECTION]: The
SAFETY.mdfile includes examples of adversarial prompt patterns (such as using[SYSTEM]tags to simulate administrative overrides) as part of a documented safety evaluation. These are used to benchmark the agent's resistance to manipulation and do not represent malicious instructions. - [DATA_EXFILTRATION]: The skill implements strict prohibitions against generating automation that transmits AWS configuration, architecture, or cost data to external APIs, webhooks, or cross-account storage buckets.
- [CREDENTIALS_UNSAFE]: Instructions explicitly forbid the agent from echoing realistic AWS credentials and mandate the use of clearly artificial placeholders (e.g.,
YOUR_SECRET_KEY_HERE) in all generated output. - [EXTERNAL_DOWNLOADS]: The skill references documentation and configuration tools hosted on the author's official vendor domain (
faberlens.ai) for security reports and guardrail management.
Audit Metadata