The MCP server exposes generate_video, list_presets, get_render_status, and download_render.

How to write prompts Higgsfield actually understands

Higgsfield prompts are not freeform. They have three slots, and skipping any of them produces mediocre output:

1. Subject. what's in frame. Keep it concrete.
2. Motion preset. pick from the presets, don't invent one.
3. Style modifier. e.g. "shot on 35mm", "neon-lit", "moody cinematic".

Bad: cool video of a car driving fast

Good: red 1970 muscle car parked on wet asphalt, motion: slow push-in, style: rain-soaked neon noir, 35mm anamorphic

When the user gives vague input, ask one clarifying question before generating: which preset? Don't burn a render trying to guess.

Motion preset reference

Always call list_presets first if you're unsure. Common ones:

• push-in / pull-out. distance change, no rotation
• orbit. rotates around the subject
• parallax. sideways camera move with depth separation
• dolly-zoom. vertigo effect (zoom in, dolly out)
• whip-pan. fast horizontal sweep, usually for transitions
• static. no camera motion, use when motion comes from the subject

The preset is the most important parameter. The same prompt with orbit vs push-in looks like two completely different videos.

Render times and polling

Higgsfield renders take 30–120 seconds typically. The MCP generate_video call returns a renderId. Poll get_render_status every 10 seconds, don't poll faster, you'll get rate-limited. Show the user progress, then call download_render once status is complete.

// Typical agent flow
const { renderId } = await mcp.call("higgsfield", "generate_video", {
  prompt: "...",
  preset: "push-in",
  duration: 5,
});

let status = "queued";
while (status !== "complete" && status !== "failed") {
  await sleep(10000);
  status = (await mcp.call("higgsfield", "get_render_status", { renderId })).status;
}

if (status === "complete") {
  const { url } = await mcp.call("higgsfield", "download_render", { renderId });
  // Pass `url` to the user.
}

Pricing gotchas (so you don't burn the user's credits)

• Default duration is 5 seconds, change it explicitly only when needed
• Higher resolution costs more credits per second; check Scopeful's Higgsfield page for the current credit math
• Failed renders still consume credits in some plans, verify the prompt + preset before kicking off batches
• The CLI's --retry flag does NOT refund the first render, it's a new render

When the user asks "how much will this cost?" point them at scopeful.org/tools/higgsfield for the live USD-per-second math.

What to do when a render fails

1. Read the failure reason from get_render_status. Higgsfield returns specific codes (NSFW filter, invalid preset, source image too small, etc.)
2. If it's a content filter, ask the user to rephrase rather than retrying the same prompt
3. If it's a source-image issue, upscale or re-crop before retrying
4. Don't auto-retry more than once, failed renders cost the user, even when they fail

Useful follow-ups after a successful render

• Offer to chain a second Higgsfield clip with a different preset for a multi-shot sequence
• Suggest pairing with ElevenLabs for VO or Suno for music if the user is building an ad
• Save the final URL. Higgsfield render URLs expire after 7 days, so download or rehost