The Agent Skills Directory

[SAFE]: No malicious patterns detected across the skill's instructions or reference documentation.
[CREDENTIALS_SAFE]: The skill provides explicit security guidance for handling sensitive data. It instructs the agent to use --password-stdin for Docker logins to avoid leaking secrets in process lists and specifies restricted file permissions (0600) for private keys when configuring custom domains.
[EXTERNAL_DOWNLOADS]: The skill references downloads from trusted or well-known sources, such as Hugging Face for model weights and the vendor's own official repositories (beclab/apps) or Docker Hub images (beclab/*). These are documented neutrally as part of the skill's primary function.
[PRIVILEGE_ESCALATION]: While the skill mentions privileged: true and runAsUser: 0 for certain Kubernetes templates, these configurations are strictly limited to trusted beclab/ images and serve legitimate purposes like volume initialization and Docker-in-Docker sidecars for development environments.
[COMMAND_EXECUTION]: Shell commands involving olares-cli, docker, and gh are used for their intended purposes in a development and deployment workflow. There is no evidence of arbitrary or malicious command execution.

olares-chart