How to Connect Higgsfield to Gemini via MCP

💡 No install? Use cloud.anythingmcp.com directly. Sign in, click Connectors → Higgsfield, paste your credentials, mint an MCP API key and you're done. No Docker, no git clone, no local server.

Higgsfield + Gemini

Let your AI create images and videos on Higgsfield with your own API key: SOUL portraits, Marketing Studio and Ads Studio product images, Seedance, Kling 3.0 and Cinema Studio videos, and 80+ other models, with price estimates before each run, status polling and cancel.

Prerequisites

See the full setup instructions baked into the connector (visible in the in-app store when you select the connector). The required environment variables for this connector are:

HIGGSFIELD_API_KEY_ID, HIGGSFIELD_API_KEY_SECRET

Step 1 — Get credentials

Setting up takes about 3 minutes. You need a Higgsfield Console account with API credits and an API key. Every generation is billed to that balance.

1. Sign in at https://console.higgsfield.ai and top up the API balance (pay as you go). Without credits, generations are refused with 403.
2. In the Console, open API keys and create a key. It has two parts: a key ID and a secret. The secret is shown only once, so copy it right away.
3. Paste the key ID into HIGGSFIELD_API_KEY_ID and the secret into HIGGSFIELD_API_KEY_SECRET, each on its own, without a colon or the word Key. The connector joins them into the header Authorization: Key <id>:<secret> and sends it only to api.higgsfield.ai.
4. Install the connector. The install checks the key by listing the SOUL styles, which costs nothing; you can repeat that check any time with higgsfield_list_soul_styles.

What it creates: images with SOUL V2 (photorealistic portraits, fashion and editorial, optionally in a SOUL style or with one of your trained Soul ID characters), Marketing Studio (campaign and product images, edits with up to 16 reference images, or ads from a preset) and Ads Studio (a brief and optional product photos become 1 to 8 ad concepts). Videos with Seedance 2.5 (from text, from an image or from reference media), Kling 3.0 (Standard, Pro or 4K tier, with native sound) and Cinema Studio 4.0. Any other of the 80+ models, such as Wan, MiniMax, Hailuo, Grok Imagine, Recraft or Ideogram, runs with higgsfield_generate by its endpoint ID from the model's page at docs.higgsfield.ai.

What it reads, for free: the status and output of a request, price estimates, the SOUL styles, the Marketing Studio presets and your Soul ID characters with their training status.

Every generation spends credits, so the AI estimates first and asks: before higgsfield_soul_image, higgsfield_marketing_studio_image, higgsfield_ads_studio, higgsfield_seedance_text_to_video, higgsfield_seedance_image_to_video, higgsfield_seedance_reference_to_video, higgsfield_kling_text_to_video, higgsfield_kling_image_to_video, higgsfield_cinema_studio_video or higgsfield_generate, the AI runs higgsfield_estimate_cost with the same model and input, tells you which model will run and what it costs, and waits for your yes. It never starts several generations in a loop without asking. Video costs far more than images and grows with duration, resolution and tier; four images in one batch cost about four times one. Failed requests and requests rejected by moderation (nsfw) are not charged. higgsfield_cancel_request also asks first.

Switched off at install: nothing. Every paid tool goes through the price estimate and your confirmation, and cancelling only works on requests that have not started.

From request to result: a generation answers at once with a request_id and the status queued. The AI checks higgsfield_get_request_status every 5 to 10 seconds; images take seconds, videos usually minutes. You get links to the images, the video or the audio, not files. The links stay available for at least seven days, so download what you want to keep. A request can be cancelled only while it is still queued, and its credits are refunded; once it is in progress, it is billed when it completes. After a timeout, a retry with the same idempotency_key returns the original request instead of a second, billed one.

Inputs are links: reference images, videos and audio go in as public HTTPS URLs. The connector cannot upload files from your computer.

Limits: the main limit is concurrency, the number of requests queued or running at the same time (often 4). A 400 saying the maximum number of concurrent requests has been reached means waiting until one finishes. Your account's limits are shown in the Console.

Common errors: 401 Invalid credentials means the key ID or secret is wrong, swapped or revoked: create a new key in the Console. 403 means not enough credits. 404 means the request id does not exist or belongs to another account, or the model is not available to your account. 422 means an input field is missing, misspelled or out of range, and the error names it. 423 or 503 means the model is temporarily blocked or disabled: try later or pick another model.

The connector does not upload files, receive webhooks, train Soul ID characters, use the Agent API, or see the projects and history of the Higgsfield web app.

Step 2 — Install the adapter

curl -fsSL https://raw.githubusercontent.com/HelpCode-ai/anythingmcp/main/docker-compose.quickstart.yml -o docker-compose.yml
printf 'JWT_SECRET=%s\nENCRYPTION_KEY=%s\n' "$(openssl rand -hex 32)" "$(openssl rand -hex 32)" > .env
docker compose up -d

Step 3 — Add the connector in Gemini

Your server URL: in AnythingMCP, open MCP Servers → the server this connector is on, and copy its URL (https://cloud.anythingmcp.com/mcp/…). Use it wherever this guide shows YOUR_SERVER_ID.

Gemini CLI reads MCP servers from ~/.gemini/settings.json (on Windows %USERPROFILE%\.gemini\settings.json). Add:

{
  "mcpServers": {
    "anythingmcp": {
      "httpUrl": "https://cloud.anythingmcp.com/mcp/YOUR_SERVER_ID",
      "headers": { "X-API-Key": "YOUR_MCP_API_KEY" }
    }
  }
}

1. Create an MCP API key in AnythingMCP under MCP Servers → your server → API keys, and paste it in place of YOUR_MCP_API_KEY.
2. Save the file and restart gemini.
3. Run /mcp inside Gemini CLI (or gemini mcp list in your shell): anythingmcp should show as connected.
4. Start prompting: Gemini CLI can now call the Higgsfield tools.

Available tools

Example prompts

• "Create a photorealistic portrait of a woman in a linen suit on a rooftop at golden hour with SOUL. Tell me the price before you start."
• "Here is a photo of our new running shoe: https://example.com/shoe.jpg. Turn it into 4 ad concepts for an Instagram campaign."
• "Make a short Kling 3.0 Pro video of a barista pouring latte art in slow motion. What will it cost?"
• "Animate this product shot into a short video with Seedance: https://example.com/bottle.jpg. Show me the estimate first."
• "Is my last video finished yet? Send me the link so I can download it."

FAQ

Where does Gemini CLI look for MCP servers? In ~/.gemini/settings.json for your user, or in .gemini/settings.json inside a project. gemini mcp list shows each server and whether it connects.

Next steps

• Connector landing page
• Connect to Claude
• Connect to ChatGPT