The Agent Skills Directory

[COMMAND_EXECUTION]: The skill makes extensive use of ssh to execute administrative commands on remote Home Assistant instances. It orchestrates complex multi-step pipelines involving the ha CLI, docker, and filesystem operations.
[PRIVILEGE_ESCALATION]: The instructions explicitly guide the user to disable 'Protection mode' on the SSH add-on to grant the agent access to the Docker socket. Furthermore, it utilizes sudo for container management and inspection, which bypasses standard permission boundaries.
[DYNAMIC_EXECUTION]: The skill uses heredocs to inject and execute dynamic Bash and Python scripts directly into remote shells. This includes a logic for merging JSON configurations where a Python script is generated at runtime to handle sensitive option objects.
[INDIRECT_PROMPT_INJECTION]: The skill is susceptible to command injection via the interpolation of user-supplied placeholders like <alias> and <slug> into shell commands.
Ingestion points: User-provided placeholders <alias>, <slug>, and configuration options defined in SKILL.md.
Boundary markers: None; placeholders are directly embedded into shell strings.
Capability inventory: Full shell access via ssh, sudo docker operations, and ha CLI interactions across all sections of SKILL.md.
Sanitization: No validation or escaping is applied to placeholders before they are interpolated into command strings.

haos-addon-deploy