The Agent Skills Directory

[COMMAND_EXECUTION]: The deployment script scripts/deploy-hermes-agent.py uses the subprocess module to execute SSH and SCP commands. These commands facilitate the transfer of configuration files and the remote management of Docker containers on the target VM. The use of list-based arguments for local execution mitigates shell injection risks.
[PRIVILEGE_ESCALATION]: The scripts/compose/compose.yaml configuration mounts the host's Docker socket (/var/run/docker.sock) into the web-terminal container. This grants the terminal user root-equivalent access to the host VM, allowing them to manage the agent container. The skill documentation explicitly identifies this as a design choice for the developer environment.
[COMMAND_EXECUTION]: The skill executes shell commands with sudo privileges on the remote VM to create data directories and start the Docker Compose stack. This level of access is appropriate for a deployment and system administration task.
[SAFE]: The skill implements secure secret handling by generating high-entropy passwords using secrets.token_urlsafe and restricting local report file permissions to the owner (mode 600).

cloud-weaver-hermes-agent