The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses a wide array of standard DevOps tools including kubectl, docker, terraform, gh, aws, trivy, tfsec, and infracost. These are used for environment detection, validation, and diagnostics consistent with the skill's purpose.
[DATA_EXFILTRATION]: The skill attempts to detect cloud environments by reading environment variables using env | grep -E 'AWS|AZURE|GOOGLE|GCP' | head -3. While used for identification, this command can expose environment variable names and potentially values (such as IDs or regions) to the agent's context. This is handled as part of the tool's diagnostic routine.
[PROMPT_INJECTION]: The skill possesses a surface for indirect prompt injection through log analysis.
Ingestion points: The skill reads logs and tool outputs from sources including gh run view <run-id> --log, kubectl logs, and docker logs, which may contain untrusted data from build processes or external applications.
Boundary markers: Absent; no specific instructions are provided to the agent to separate log content from internal instructions.
Capability inventory: The skill has extensive execution capabilities including shell access, container management (docker), cluster management (kubectl), and infrastructure deployment (terraform).
Sanitization: Absent; content from external logs is processed directly without explicit filtering or validation steps.

devops-expert