The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses several system commands for environment detection and configuration. These include uname -a, systemctl, docker, and chmod 600 on environment files. These are standard system administration tasks required for the stated onboarding purpose.
[PERSISTENCE]: The skill configures systemd services for the Hermes gateway and dashboard to ensure they remain active across system restarts. It uses standard unit file locations like /etc/systemd/system/ and ~/.config/systemd/user/.
[DATA_EXPOSURE_AND_EXFILTRATION]: The skill instructs the agent to store bot tokens in ~/.hermes/.env and explicitly warns against pasting tokens into chat transcripts. It recommends binding the dashboard to 127.0.0.1 (loopback) and accessing it via SSH tunnels, which is a secure practice.
[REMOTE_CODE_EXECUTION]: The skill mentions installing plugins and skills from the moonlight-lupin/agent-skills repository. As this repository belongs to the skill's author, this is considered expected vendor functionality for extending the agent's capabilities.
[SAFE]: The weekly_health_check.sh script performs local diagnostics (CPU load, disk usage, log scanning). It does not perform network operations or execute remote code; it is a passive monitoring tool.

hermes-onboarding