infra-containers-kubernetes

Pass

Audited by Gen Agent Trust Hub on Jul 25, 2026

Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTIONEXTERNAL_DOWNLOADSCREDENTIALS_UNSAFE
Full Analysis
  • [PROMPT_INJECTION]: The skill processes user-supplied configuration requirements to generate Kubernetes manifests, creating a surface for indirect prompt injection.
  • Ingestion points: User requests for workload specifications in SKILL.md.
  • Boundary markers: Uses markdown headers and structured sections to separate instructions.
  • Capability inventory: Provides kubectl commands for cluster interaction in reference.md.
  • Sanitization: Mandates restricted security contexts (non-root, read-only FS) which serve as a security control.
  • [COMMAND_EXECUTION]: The reference.md file contains kubectl commands such as apply, exec, and logs for managing cluster resources.
  • [EXTERNAL_DOWNLOADS]: Fetches Helm chart dependencies from the well-known Bitnami repository as seen in examples/helm.md.
  • [CREDENTIALS_UNSAFE]: Includes illustrative dummy credentials in manifest examples, such as a placeholder Stripe key 'sk-live-abc123' and generic database connection strings in examples/core.md and SKILL.md.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 25, 2026, 02:34 AM
Security Audit — agent-trust-hub — infra-containers-kubernetes