The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes standard system and tool-specific commands to verify the user's environment and demonstrate core functionality.
Runs kubectl cluster-info and kubectl get pods to verify cluster access.
Uses mirrord ls and mirrord exec to manage local development sessions connected to Kubernetes.
These commands are standard for the tool's documented purpose and do not represent a security risk.
[INDIRECT_PROMPT_INJECTION]: The skill processes output from the Kubernetes cluster (pod names, cluster info) and the mirrord CLI.
Ingestion points: kubectl get pods -A, kubectl cluster-info, and mirrord ls (SKILL.md).
Boundary markers: None present in the command blocks.
Capability inventory: Execution of kubectl and mirrord CLI tools.
Sanitization: Not explicitly performed on command output before agent processing. While pod names or cluster metadata could theoretically contain malicious instructions, the risk is negligible for this use case.
[SAFE]: The skill includes security-conscious instructions, explicitly warning the user and the agent not to use one-liners that fetch and execute remote scripts (e.g., curl | bash). It correctly points to official documentation for installation steps.

mirrord-quickstart