The Agent Skills Directory

Indirect Prompt Injection Surface: The skill analyzes data from the cluster environment, such as Helm manifests and EKS Insights, which could be exploited to influence the agent's reasoning. * Ingestion points: Helm secrets (Step 4), EKS Insights API (Step 3), and resource metadata (Step 6, Step 9). * Boundary markers: The skill uses a structured 17-step workflow and explicit status gates (e.g., "UNKNOWN ≠ PASS") which provide logical boundaries, though no specific delimiters are defined for data interpolation. * Capability inventory: The skill is limited to read-only AWS CLI and kubectl commands and does not have the ability to perform network exfiltration or execute arbitrary code. * Sanitization: Data extracted from Helm manifests is analyzed for API versions without explicit sanitization of non-relevant content.
Extensive Command Execution: The skill relies on a wide array of AWS CLI and kubectl commands to perform its assessment. While these operations are read-only, they involve high-privilege access to gather detailed infrastructure and workload metadata.
External Resource References: The skill documentation includes links to official AWS documentation and well-known third-party tools (e.g., kubent, pluto). These resources are provided for informational purposes for the operator and are not automatically executed by the skill.

eks-upgrade-readiness