Claude AI smart home integration, free and open source
Run the Gladys assistant on Claude, the AI of Anthropic, with your own API key.
Published by prohand
- Devices
- Version 0.1.1
- Requires Gladys >=5.3.0
- ★ 0
This integration makes Claude, the AI of Anthropic, the brain of the Gladys assistant. Once selected, it answers every AI request of your Gladys: the chat (and the devices it controls for you), the routing of your messages, the Ask the AI scene action and the weekly digest.
It is a community integration: it is not made nor supported by Anthropic.
Before you start
- Gladys 5.3 or later.
- An Anthropic API key. Create an account on
console.anthropic.com, add credit or a
payment method, then create a key under Settings → API keys. It starts
with
sk-ant-. - Usage is billed by Anthropic, per token, at the price of the model you choose. Gladys Plus is not needed.
Claude runs in the cloud (api.anthropic.com): Gladys needs internet access.
Configuration
- Install the integration from the catalog, then open its Configuration tab.
- Paste your Anthropic API key and save.
- Click Test the connection: each model answers a one-word question and the time it took is shown under the button.
- Go to Settings → Artificial Intelligence and select Claude AI as the AI provider.
The status at the top of the Configuration screen turns green when the key is accepted and the chosen models are open to it. When it is red, it says why.
The settings
| Setting | What it does |
|---|---|
| Anthropic API key | Your key. It stays in Gladys, is never shown again and never written to the logs. |
| Chat model | Answers the chat, creates the scenes, writes the weekly digest. Claude Opus 5.5 by default. Sonnet 5.5 is faster and cheaper, Fable 5.1 the strongest and the most expensive. |
| Router model | Sorts each message before the answer. Gladys waits 20 seconds at most for it: keep a fast model, Claude Haiku 5.5 by default. |
| Reasoning effort | How much the chat model thinks before it answers: low, medium (default), high, very high. More effort helps on hard requests (creating a scene) but answers slower and costs more. |
| Maximum answer length (tokens) | The output budget of one answer, reasoning included. 16,000 by default. Raise it if the logs say an answer hit max_tokens. |
What Claude sees
Choosing an AI provider is a privacy decision. The provider receives:
- every chat message and the conversation history;
- the description of the tools Gladys offers it, and their results — the state of the devices it reads;
- the images you send in the chat.
Its answers drive the control of your devices and the creation of your scenes. Read the Anthropic privacy policy before you choose it.
Costs and speed
- The system prompt and the tool list come back on every message: the integration asks Anthropic to cache them, which bills them at a fraction of the price after the first message.
- The router always runs at low effort on the router model: it is the request sent the most often.
- To spend less, pick Claude Sonnet 5.5 for the chat or lower the effort.
- When Claude declines a request for safety reasons, Anthropic retries it on its default fallback model inside the same call; if every model declines, the assistant says it cannot help.
Troubleshooting
- "Anthropic refused the API key" — the key is wrong, revoked or belongs to an organization without credit. Create a new one and paste it again.
- "… is not available to this API key" — your Anthropic organization has no access to that model: choose another one.
- "Rate limit reached" — too many requests for your Anthropic usage tier: wait a moment, or raise your tier in the Anthropic console.
- "Cannot reach api.anthropic.com" — Gladys has no internet access, or a firewall blocks that address.
- The assistant says the AI provider is unavailable — open the Logs tab of the integration: every request writes one line (model, time, tokens) and every failure its reason.
Configuration settings
These are the settings Claude AI asks for in its configuration screen in Gladys.
| Setting | Type | Required | Description |
|---|---|---|---|
| Getting started | section | No | Claude answers through the Anthropic API, in the cloud: create an API key on console.anthropic.com (usage is billed by Anthropic), paste it below, then select this integration in Settings, Artificial Intelligence. Claude then sees every chat message, the history, the state of the devices it reads and the chat images. |
| Anthropic API key | secret | Yes | Created on console.anthropic.com, it starts with sk-ant-. It stays in Gladys and is never shown again nor logged. |
| Chat model | select | Yes | Answers the chat, creates the scenes and writes the weekly digest. Opus is the most capable for its price, Sonnet faster and cheaper, Fable the strongest and the most expensive. |
| Router model | select | Yes | Sorts every chat message before the answer, and must reply within 20 seconds: a fast model fits best. |
| Reasoning effort | select | No | How much the chat model thinks before it answers. More effort: better answers on hard requests, but slower and more tokens billed. |
| Maximum answer length (tokens) | number | No | The reasoning counts in it too. Raise it if the logs say an answer hit max_tokens. |
How to install Claude AI in Gladys
- In Gladys, open Integrations: Claude AI appears in the catalog, next to the native integrations, with a community badge.
- Click Install. Gladys pulls the Docker image (
ghcr.io/prohand/gladys-ai-claude:0.1.1), starts it in a sandbox isolated from the core, and generates the integration's interface (devices, discovery and configuration). - Open the Configuration screen of the integration, fill in the settings, and save.
- You can also install it directly from its repository URL: https://github.com/prohand/gladys-ai-claude.
Claude AI requires Gladys >=5.3.0. The catalog inside Gladys refreshes every hour, so a new version becomes available at most one hour after its release.
Not running Gladys yet? It is free and open source: follow the installation guide to get started.
About external integrations
Claude AI is an external integration: a community integration packaged as a Docker container and published on GitHub, that Gladys installs in one click and runs in a sandbox isolated from its core. It is published and maintained by prohand, not by the Gladys core team.
- Browse all integrations, native and community
- Build and publish your own external integration
- Source code on GitHub — source of this documentation