The Agent Skills Directory

[COMMAND_EXECUTION]: The skill documents the uc service exec command, which allows the agent or user to execute arbitrary shell commands inside running containers within the cluster.
[COMMAND_EXECUTION]: The instructions include commands for cluster initialization and machine management (uc machine init, uc machine add), which involve bootstrapping remote hosts, typically requiring administrative privileges.
[EXTERNAL_DOWNLOADS]: The skill references container images from trusted and well-known registries, such as registry.k8s.io (Kubernetes official registry) and standard Docker Hub images (e.g., nginx, postgres).
[PROMPT_INJECTION]: The skill describes a processing workflow for external data that creates a surface for indirect prompt injection.
Ingestion points: The skill processes compose.yaml files and custom Caddyfile snippets provided by the user.
Boundary markers: Absent; there are no specific delimiters or instructions for the agent to ignore potentially malicious embedded instructions in these files.
Capability inventory: The agent has access to powerful cluster operations, including service deployment (uc deploy), container execution (uc service run), and remote shell access (uc service exec).
Sanitization: Absent; the skill does not detail any validation, escaping, or sanitization requirements for the processed configuration content.

uncloud