The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill directs the agent to ingest data from external sources that are not fully under the agent's control, such as pod logs and cluster events, creating a surface for indirect instructions to influence agent behavior.
Ingestion points: oc logs -f pod/myapp-1-xyz (SKILL.md), oc get events --sort-by='.lastTimestamp' (SKILL.md).
Boundary markers: The instructions lack delimiters or specific warnings to treat ingested log/event content as data only and to ignore any instructions embedded within them.
Capability inventory: The skill utilizes the powerful oc CLI which supports resource creation (oc apply), build execution (oc start-build), and interactive debugging (oc debug), which could be targeted by injected instructions.
Sanitization: No sanitization, escaping, or validation of log or event content is mentioned before processing.
[COMMAND_EXECUTION]: The skill uses the oc (OpenShift CLI) to perform administrative tasks. While standard for DevOps, these commands have significant impact on the target environment.
Commands such as oc debug pod/... and oc start-build --from-dir=. involve executing code or interactive sessions that interact with local and remote filesystems.
The skill provides authentication examples using oc login with token and server placeholders.

openshift