The Agent Skills Directory

[COMMAND_EXECUTION]: The skill provides instructions for performing high-privilege system tasks using sudo, including managing the crio service via systemctl, editing system configuration files in /etc/crio/, and executing interactive shell commands within containers using crictl exec. This functionality is consistent with the skill's primary purpose of container runtime administration.
[EXTERNAL_DOWNLOADS]: The references/install-and-setup.md documentation includes commands to download repository keys and software binaries from pkgs.k8s.io and github.com. These sources are recognized as official and well-known providers for the Kubernetes and container networking ecosystems.
[PROMPT_INJECTION]: An indirect prompt injection surface is identified as the skill enables the agent to read and process potentially untrusted data from the system environment.
Ingestion points: The agent is instructed to use crictl logs and crictl inspect to retrieve container output and metadata, as described in references/command-cookbook.md.
Boundary markers: The instructions do not specify the use of delimiters or provide explicit warnings to the agent to disregard instructions embedded within log content.
Capability inventory: The agent has access to administrative capabilities, including configuration modification (sudo vi), service restarts, and command execution within containers (crictl exec), as outlined in SKILL.md and references/crio-configuration.md.
Sanitization: No sanitization or validation of the data retrieved from tool outputs is included in the workflows.

cri-o