The Agent Skills Directory

[COMMAND_EXECUTION] (SAFE): The skill utilizes kubectl exec to interact with the OpenBao pod for initialization, unsealing, and secret management. These are standard administrative operations for Kubernetes-based secret vault management and are consistent with the skill's intended purpose.
[CREDENTIALS_UNSAFE] (SAFE): The provided examples for secret storage (e.g., username=myuser, password=mypassword) and placeholders (e.g., <key1>) are generic and do not contain real credentials or sensitive keys.
[DATA_EXFILTRATION] (SAFE): No unauthorized network operations or file access patterns were identified. The skill focuses on synchronizing data within the Kubernetes cluster boundary.
[INDIRECT_PROMPT_INJECTION] (LOW): The skill describes a surface where external data (secrets from OpenBao) is ingested and interpolated into templates. 1. Ingestion points: ExternalSecret resource in SKILL.md. 2. Boundary markers: YAML structure. 3. Capability inventory: kubectl subprocess calls in SKILL.md. 4. Sanitization: Absent. While this is an attack surface, it is a standard implementation of the External Secrets Operator and does not indicate malicious intent.

secrets-management