The Agent Skills Directory

[CREDENTIALS_UNSAFE]: The skill documentation contains hardcoded credential examples within the Secret YAML section, specifically using Base64-encoded strings that decode to 'admin' and 'secret123'. While intended as placeholders, they represent hardcoded sensitive patterns.
[DATA_EXFILTRATION]: The skill demonstrates how to create Kubernetes secrets from sensitive local file paths, specifically providing a command example that reads the user's SSH private key from ~/.ssh/id_rsa.
[INDIRECT_PROMPT_INJECTION]: The skill provides resource templates that define an ingestion surface for untrusted external data into a cluster environment. 1. Ingestion points: Data fields and environment variable references in the ConfigMap and Secret templates within SKILL.md. 2. Boundary markers: Absent; the templates do not include delimiters or instructions to ignore potential commands embedded in the configuration values. 3. Capability inventory: The skill utilizes the kubectl CLI to create resources and includes instructions for using kubectl exec to inspect file systems and environment variables within running containers. 4. Sanitization: Absent; the templates do not implement validation or filtering for the interpolated configuration data.

configmap-secret