The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill downloads multiple CLI tools and utilities including KooCLI, kubectl, velero, obsutil, image-migrator, and k8clone. All downloads originate from official domains of well-known organizations (Huawei Cloud, Kubernetes, VMware) or trusted service providers (GitHub).
[DYNAMIC_EXECUTION]: The shell script scripts/test-cli-commands.sh employs the eval command to execute shell commands extracted from templates/test-vars.json. This pattern involves dynamic assembly and execution of code based on externalized configuration files.
[INDIRECT_PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection due to its core functionality of reading state from source clusters and applying it to target clusters. It ingests potentially untrusted data from resource manifests and metadata which could influence the agent's behavior during the migration process.
Ingestion points: Commands like kubectl get and hcloud list operations in references/pre-migration-checklist.md and references/external-dependency-audit-guide.md extract workload data into the agent's context.
Boundary markers: No explicit delimiter or instruction to ignore embedded directives is used when interpolating cluster data into command strings.
Capability inventory: The skill uses kubectl patch, kubectl apply, and various hcloud administrative commands to modify cluster state, provision storage, and configure networking.
Sanitization: The migration guides (e.g., references/k8clone-migration-guide.md) specify manual and automated sanitization rules (stripping UIDs, resourceVersions, and excluding system secrets), which serves as a mitigation factor for accidental data corruption or injection.
[COMMAND_EXECUTION]: The skill performs extensive command execution using administrative tools (hcloud, kubectl, velero) to manage cloud infrastructure and Kubernetes resources. This behavior is expected and necessary for its primary purpose as a migration assistant.

huawei-cloud-cce-cluster-migrator