The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The documentation in SKILL.md provides instructions to download and execute an installation script from https://fluxcd.io/install.sh. This is an official source for the FluxCD tool.
[COMMAND_EXECUTION]: The skill enables the execution of Kubernetes management commands through kubectl and flux. This includes high-privilege operations such as kubectl exec, which allows the agent to run arbitrary commands inside containerized environments.
[PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection via untrusted data retrieved from the cluster.
Ingestion points: Data enters the agent context through commands like kubectl logs, kubectl get events, and flux get all as described in SKILL.md and implemented in scripts/main.cjs.
Boundary markers: No specific delimiters or instructions to ignore embedded commands are present in the scripts or system prompts.
Capability inventory: The skill possesses the Bash tool and utilizes child_process.spawn in scripts/main.cjs to run CLI tools. It supports kubectl exec for remote command execution within pods.
Sanitization: There is no evidence of sanitization, filtering, or validation of the retrieved log data or event messages before they are processed by the agent.

kubernetes-flux