The Agent Skills Directory

[DATA_EXFILTRATION]: The skill uses kubectl exec -n crs <pod> -- env to display environment variables. In containerized environments, sensitive information such as API keys, service credentials, or internal configuration is frequently stored in environment variables, which can lead to unintentional data exposure.
[PROMPT_INJECTION]: The skill processes untrusted data from container logs (kubectl logs) and Kubernetes events (kubectl get events), which represents a surface for indirect prompt injection.
Ingestion points: External data enters the agent context via container logs, Kubernetes events, and command outputs from within pods as defined in SKILL.md.
Boundary markers: None. The skill does not provide delimiters or instructions to the agent to treat the ingested data as untrusted or to disregard embedded commands.
Capability inventory: The skill has the capability to interact with the Kubernetes cluster using kubectl and execute shell scripts using bash.
Sanitization: There is no evidence of sanitization, filtering, or validation of the log and event content before it is processed by the agent.
[COMMAND_EXECUTION]: The skill executes local bash scripts deployment/collect-logs.sh and scripts/diagnose.sh. The latter utilizes a template variable {baseDir}, which indicates dynamic path resolution for script execution at runtime.

debug-buttercup