The Agent Skills Directory

[CREDENTIALS_UNSAFE]: The skill contains hardcoded default credentials for local development services.
Evidence: KEYCLOAK_ADMIN_PASSWORD=admin and POSTGRES_PASSWORD=keycloak are defined in the docker-compose.yaml section.
Evidence: client-secret=local-dev-secret and AWS mock keys (accessKeyId: "test", secretAccessKey: "test") are used in the local configuration and scripts.
Note: These credentials are used for local sandboxed containers and appear intended for demonstration or development purposes rather than accessing sensitive user accounts.
[EXTERNAL_DOWNLOADS]: The skill downloads and executes Kubernetes configuration manifests from an external source.
Evidence: The scripts/dev-up.sh file executes kubectl apply -f https://raw.githubusercontent.com/kubernetes/ingress-nginx/main/deploy/static/provider/kind/deploy.yaml.
Source: This is an official repository from the Kubernetes organization, which is a well-known and trusted technology project.
[INDIRECT_PROMPT_INJECTION]: The skill processes external data which could theoretically serve as an injection vector.
Ingestion points: Remote Kubernetes manifests are downloaded and applied to the local cluster in scripts/dev-up.sh.
Boundary markers: No explicit delimiters or warnings are present in the scripts to prevent the interpretation of instructions within the downloaded manifests.
Capability inventory: The skill has the capability to execute shell commands (Bash), manage local Docker containers, and modify the local filesystem.
Sanitization: No sanitization or integrity checks (such as checksum verification) are performed on the downloaded manifests before execution.

local-eks-development