You can connect Higgsfield to Claude and then generate images, video and audio straight from the conversation. Your Soul characters are available there just as they are in the Higgsfield interface. The setup takes about a minute.
Last updated:
You need an active paid Higgsfield subscription and a Claude account. There is no API key to create, and you do not have to write any code.
Every generation you run through Claude takes credits off your Higgsfield account, no matter which plan you are on. Going through the chat is more convenient, but it is not cheaper than the interface.
In the terminal it takes a single command, which you can copy below. The sign-in then runs through the browser in the same way.
claude mcp add --transport http --scope user higgsfield https://mcp.higgsfield.ai/mcp
Open a new conversation and ask this. If Claude gives you a real number, the connection is live. If it answers in general terms without touching your account, the connector is not active yet.
Show me my current Higgsfield credit balance and my last five generations.
From here you work in the chat the way you normally would in the interface. The difference is that you can talk to Claude about the result and have it adjusted right away.
Generate an image with my Soul character. Subject: [what should be visible] Background: [where] Lighting: [e.g. soft daylight, overcast sky] Format: [9:16 or 4:5] The face should be razor-sharp, with hyper-realistic skin texture and visible pores. Before you start, show me the prompt you are about to send so I can still correct it.
Take the image from the last generation and turn it into a video. Motion: [what should happen] Camera: [static or a slow move] Length: 6 seconds Format: 9:16 Tell me how many credits this will cost before you run it.
Both prompts end by asking Claude to show you what it is about to send. Without that line it starts straight away, and you pay for attempts you never wanted.