🛡️ Hass Guard

Limits the agents behind Hassonic and the Octo MCP (hassonic and external) to an allowlist of Home Assistant actions, and blocks any action their decision model doesn’t rate safe.

Configuration Schema

FieldTypeRequiredDescription
agentsstring[]OptionalAgent IDs to guard. Defaults to ["hassonic"].
alertTargetstringOptionalWhere to post blocked calls, as <channel>:<target>, e.g. "discord:channel:<id>". Unset means no alerts.

Example config

"plugins": { "entries": { "hass-guard": { "enabled": true, "config": { "agents": ["hassonic", "external"], "alertTarget": "discord:channel:<id>" } } } }

The safety check uses each guarded agent’s own decisionModel.