Anthropic (Claude)
Anthropic makes the Claude family of models. Connect your Anthropic key and the agent can build Claude-powered features into the apps you make with Appsanic: a chat assistant, a "summarise this" button, a classifier, or anything that turns natural language into structured JSON your app can act on.
On the marketing site this connector is shown as Claude. The provider company is Anthropic, and the key you paste in comes from the Anthropic Console.
What you get by connecting it
With your key connected, the agent can build features that call Claude for:
- Reasoning and chat - in-app assistants, question answering, and multi-turn conversation.
- Summarising and rewriting - condense a document, change the tone of a draft, extract the key points.
- Classification and extraction - sort free text into categories, or pull fields out of messy input.
- Structured (JSON) output - ask Claude for a shape your app can parse directly, rather than loose prose.
The first time the agent uses your connected Anthropic key in a build, it says so briefly in its opening sentence, so you always know when a Claude feature is being wired in.
Connecting it
You connect with a single API key, which starts with sk-ant-.
- Sign in to the Anthropic Console and open API keys.
- Create a key and copy it. You only see the full value once, so copy it before you close the dialog.
- In the Appsanic project build, select Connectors → Anthropic and enter the key.
The connection applies only to the project where you save it and requires a paid plan.
Connector OAuth is not currently available for any provider. Enter the key in the project's connector form. Once saved, the key is shown only as dots, never the value. Use Test now to re-check it against Anthropic or Disconnect to remove it from the project.
Keys and security
An Anthropic key is a secret - anyone holding it can spend money on your account - so it must never be bundled into an app people download. This connector is the secret key plus relay tier.
Instead of putting the key in the app, the generated app must call Claude through a server-side relay that holds the key in its own environment. The agent can generate that relay and its setup instructions, but connecting Anthropic does not deploy the relay or transfer the saved key. Deploy the backend yourself (a Supabase Edge Function is one option) and add a separate Anthropic key directly to that runtime's secret manager.
A few things hold throughout:
- Your saved key is encrypted at rest with AES-256-GCM. The browser only ever sees that a secret exists, shown as dots, never the value.
- The model prompt never receives the decrypted key. Server-side connector code may use it for validation and authorised build work, and the agent never asks you for a key in chat. If a key enters chat anyway, revoke and replace it.
- Any key that has ever touched a chat should be rotated in the Anthropic Console.
Useful links
- Anthropic's official documentation: https://docs.anthropic.com
- Appsanic guide: AI Features in Your App
Next
Read AI Features for how to design and prompt Claude-powered features in the apps you build.