eks-platform-engineering

Warn

Audited by Gen Agent Trust Hub on Jul 14, 2026

Risk Level: MEDIUMPROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
  • [PROMPT_INJECTION]: The skill uses deceptive future-dated narratives (claiming Amazon Q Developer reaches end-of-support in 2027) to override the agent's baseline knowledge and steer users toward "Kiro", a tool from an unverified domain. This is a form of instruction override using false technical context.
  • [EXTERNAL_DOWNLOADS]: The documentation promotes and directs users to an unverified third-party domain (kiro.dev) for "official" migration guides and platform generation. This domain is not an official resource of the stated vendor and poses a risk of directing users to untrusted software or documentation.
  • [PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface by instructing the agent to ingest and use configuration patterns from the deceptive "Kiro" documentation.
  • Ingestion points: Reference files provide structural templates and steering context (references/genai-platform-engineering.md).
  • Boundary markers: Absent; there are no instructions to validate the authenticity of the "Kiro" specifications against official AWS documentation.
  • Capability inventory: The skill is designed to generate complex infrastructure artifacts including CI/CD pipelines, IAM roles, and Kubernetes manifests.
  • Sanitization: No sanitization or validation logic is provided to verify the safety of generated configurations derived from external unverified specifications.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Jul 14, 2026, 11:00 AM
Security Audit — agent-trust-hub — eks-platform-engineering