The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill downloads the pup CLI binary from the official datadog-labs GitHub repository. This download is necessary for the skill's primary diagnostic functions and targets a repository owned by the skill's author.
[COMMAND_EXECUTION]: The skill executes various system and cloud-native commands, including kubectl, docker, and kind. These are used to inspect pod states, verify webhook configurations, and apply necessary fixes to the Kubernetes environment. It also manages environment variables like DD_APP_KEY for authentication with vendor services.
[INDIRECT_PROMPT_INJECTION]: The skill has an attack surface for indirect prompt injection as it ingests untrusted data from the cluster environment, such as pod specifications, labels, and event logs.
Ingestion points: Kubernetes resource manifests retrieved via kubectl get and kubectl describe in SKILL.md.
Boundary markers: The skill does not implement explicit delimiters for the data it processes.
Capability inventory: Includes the ability to apply configuration changes (kubectl apply), restart services (kubectl rollout restart), and build container images (docker build).
Sanitization: The skill does not explicitly sanitize the external content, but the impact is mitigated by instructions requiring the agent to obtain user confirmation before performing critical actions like restarts.

troubleshoot-ssi