Home Assistant Control: why it is not safe to use
We found broad state-changing Home Assistant actions without built-in confirmation, plus setup steps that fetch and install external tools.
What we found
Our review did not find hidden instructions, hardcoded secrets, privilege escalation behavior, or evidence of unrelated credential harvesting in the inspected repository files. The code appears focused on sending data to the user-configured Home Assistant instance.
The main concern is that this skill exposes a wide range of state-changing actions, including device control, scene activation, add-on install or uninstall, package install or uninstall, and automation create, update, or delete. In the inspected server code, we did not find a built-in confirmation layer for those actions. That means an agent or workflow could issue impactful changes without an extra approval step inside the skill itself.
We also found that the macOS setup script fetches and runs a remote Homebrew install command with `curl`, and can optionally install an additional Brave Search MCP package through `npm` if the user chooses that path.
What to do instead
Use this skill only with accounts and Home Assistant scopes you are comfortable letting automation change. Add your own confirmation, review, or allowlist controls before enabling write actions. Prefer manual review of setup scripts and install optional packages only if needed.
Source: https://github.com/tevonsb/homeassistant-mcp
We report what our security review found at the time we checked, with the goal of keeping people safe. Projects change; if a maintainer has since fixed this, we are glad to recheck it. Email hello@agentpod.com.