The Agent Skills Directory

[COMMAND_EXECUTION]: Scripts such as scripts/k8s_can_i_abuse.py and scripts/gcp_impersonation_mapper.py utilize the subprocess.run function to execute system commands through the kubectl and gcloud command-line interfaces.
[DATA_EXFILTRATION]: The skill includes utilities like scripts/imds_harvester.py, scripts/entra_device_code_phish.py, and scripts/oidc_trust_auditor.py designed to extract sensitive cloud credentials and cleartext secrets from link-local metadata endpoints (IMDS), authentication flows, and Terraform state files.
[PRIVILEGE_ESCALATION]: The skill provides comprehensive tooling and documentation (scripts/aws_privesc_enum.py, scripts/gcp_impersonation_mapper.py) to identify and exploit privilege escalation primitives in AWS, GCP, Azure, and Kubernetes RBAC.
[PERSISTENCE]: The skill documents methods to establish persistence within target environments, such as the creation of malicious cron jobs on container host systems (references/kubernetes-container-escape.md) and the registration of attacker-controlled devices in Microsoft Entra ID (references/azure-entra-attacks.md).
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external data from Terraform state files (scripts/oidc_trust_auditor.py) and Kubernetes configuration files (scripts/k8s_can_i_abuse.py) without implementing sanitization or boundary markers, which could allow malicious data to influence the agent's behavior. The skill maintains an inventory of high-privilege capabilities including network operations and subprocess execution.

cloud-security