A security-first smart home integration that controls Home Assistant devices using natural language and restricted API access.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install home-assistant-agent-secure
2. Manual Installation
Copy the skill folder to one of these locations
~/.openclaw/skills/ <project>/skills/ Priority: Workspace > Local > Bundled
3. Prompt Installation
Copy this prompt to OpenClaw to install it automatically.
Help me install home-assistant-agent-secure using Clawhub. If Clawhub is not installed, install it first (npm i -g clawhub).
Prefer to download?
Get the raw skill files in a ZIP archive.
What is Home Assistant Agent (Secure)?
The Home Assistant Agent (Secure) provides a robust bridge between AI agents and your smart home ecosystem. Unlike traditional integrations that require broad administrative access, this skill is built on a restricted security model that uses the Assist (Conversation) API. It allows Openclaw Skills to interact with your home using natural language while ensuring that the agent only has access to specific areas and entities you define.
By leveraging Home Assistant's built-in Natural Language Understanding (NLU), the skill minimizes token usage and maximizes privacy. It operates via a long-lived access token assigned to a non-admin user, creating a safe sandbox for home automation commands without exposing sensitive system configurations or state data.
Home Assistant Agent (Secure) Use Cases
- Controlling lighting, climate, and appliances through natural language voice or text commands.
- Managing smart home devices in shared environments where restricted user access is required for safety.
- Operating Home Assistant in multi-lingual households with support for complex grammatical inflections.
- Enhancing Openclaw Skills deployments with secure, localized IoT control capabilities.
How Home Assistant Agent (Secure) Works
- The skill receives a natural language input from the user along with a detected language code.
- It constructs a secure POST request to the /api/conversation/process endpoint of the configured Home Assistant instance.
- The request is authenticated using a restricted Long-Lived Access Token.
- Home Assistant's Assist engine processes the intent and returns a success or error response.
- If an error occurs in an inflected language (like Polish or German), the skill automatically performs a nominative retry to find the correct entity.
- The final status and speech response are returned to the user.
Home Assistant Agent (Secure) Setup
1. Create a Restricted User
In Home Assistant, go to Settings → People and add a new user (e.g., 'openclaw-bot'). Do not make them an administrator. Assign specific areas and entities to this user.
2. Generate Token
Log in as the new restricted user, navigate to your Profile, and create a Long-Lived Access Token.
3. Configure Environment
Add your credentials to your Openclaw Skills environment configuration in ~/.openclaw/.env:
HOME_ASSISTANT_URL=https://your-ha-instance.local
HOME_ASSISTANT_TOKEN=your-restricted-token
Ensure that allow_bypass_login is disabled in your Home Assistant configuration.yaml to maintain the integrity of the restricted user model.
Home Assistant Agent (Secure) Data Schema & Taxonomy
The skill utilizes a structured JSON exchange with the Home Assistant API.
| Field | Type | Description |
|---|---|---|
text |
String | The natural language command provided by the user. |
language |
String | ISO language code used for NLU processing. |
response_type |
String | Indicates the outcome (e.g., action_done or error). |
speech |
Object | Contains the verbal confirmation returned by Home Assistant. |
data |
Object | Lists the specific entity IDs and names impacted by the command. |
Home Assistant Agent (Secure) Advanced Features
- Nominative Retry: Intelligent handling of grammatically complex languages to ensure entity matching even when words are inflected.
- Restricted Security Model: Strict enforcement of the Conversation API to prevent unauthorized access to system states or configuration.
- Self-Signed Certificate Support: Includes flags for local network environments using internal SSL certificates.
- Multi-Agent Compatibility: Designed to work within the broader Openclaw Skills ecosystem for sophisticated automation workflows.
SKILL.md
Loading
Related Openclaw Skills
ATXP: AI Search & Multi-Media Suite
A comprehensive toolset for real-time web search, social media intelligence, and AI-driven media generation.
Skill Preflight Bootstrap
A one-click tool to initialize standardized AI agent pre-check workflows and learning logs in any repository.
Skill Auditor
A comprehensive security auditing tool designed to detect malicious code, obfuscated payloads, and social engineering threats in Openclaw Skills before installation.
Skill Auditor
A multi-layered security scanner designed to detect malicious code, obfuscated payloads, and social engineering threats within Openclaw Skills.
Local Piper TTS Multilang Secure
A secure, fully offline text-to-speech engine powered by Piper, featuring automatic language detection and on-demand voice model management.
ColorKit
A professional AI guide for the ColorKit Swift library, specializing in advanced color space conversion, blending modes, and accessibility compliance.
Featured*
