movie_editFLUX Video Agent in Infoseek

FLUX Video Agent

Turn a simple idea, product photo, or existing clip into a polished short video. Describe what you want in plain language and refine the result through the agent.

  • check_circle Create product spots, social clips, and cinematic scenes
  • check_circle Animate images and guide a shot through keyframes
  • check_circle Continue a good clip or make one focused edit
  • check_circle Review, save, download, and share the finished video

Prompt

A scuba diver swims slowly through a dark underwater cavern, shining a bright torch across vivid blue coral as particles drift through the water. Smooth tracking camera, deep blue light, cinematic documentary realism.

Make the brief more visual

Start with a sentence, a frame, a few ordered images, or a clip that is almost right. The agent helps turn that direction into a usable video.

Text to video

Describe the subject, action, camera, atmosphere, and sound for a new short scene. The example is a cinematic close-up of an eagle in a misty forest.

Animate images and keyframes

Bring a still image to life or guide one continuous shot through an ordered set of frames.

Continue a clip

Extend a good take while keeping its subjects, movement, camera language, and visual continuity.

Source clip

Edited result

Edit an existing video

Attach one MP4 and ask for a focused change—remove an object, add something, change a color, restyle the scene, or adjust dialogue—while keeping the rest of the take intact.

schoolFLUX prompting guide

Prompts that give the agent direction

A useful brief does not need technical language. Say what is happening, where it happens, how the camera moves, what the light feels like, and what must stay consistent.

1. Start with the action

Name the subject and the motion before adding style. The official example describes an African penguin waddling across sun-warmed granite boulders as a handheld camera follows.

Try this

“An African penguin waddles across sun-warmed granite boulders behind swaying fynbos foliage, then hops down a rock shelf and pushes on between the boulders as the handheld camera follows.”

2. Add camera and atmosphere

Use everyday camera language—slow push-in, overhead view, handheld follow, wide shot—then add lighting, weather, color, or mood. The example video follows a diver through a blue underwater cavern.

Try this

“A scuba diver swims slowly through a dark underwater cavern, shining a bright torch across vivid blue coral as particles drift through the water. Smooth tracking camera, deep blue light, cinematic documentary realism.”

3. Be precise when editing

For an existing clip, name only the change you want and say what should remain untouched. This helps the agent preserve a good take.

Try this

“Remove all the pedestrians and add a flock of pigeons around the fountain.”

helpVideo creation FAQs

Questions people ask first

What can the agent create?

New scenes from text, image animations, keyframe-led shots, continuations of existing clips, and focused edits to an MP4.

How long can a new video be?

New videos support whole-second durations from 5 to 20 seconds. Continuation uses shorter lengths, and editing keeps the source clip’s duration.

What does Video editing change?

Give the agent one MP4 up to 15 seconds and describe the change. Editing preserves the source timing and framing; anything you do not ask to change should stay as shot.

Can I make vertical social content?

Yes. Ask for 9:16 vertical video for Reels, Shorts, or TikTok, or choose a landscape or square composition for other channels.

Can I use my own media?

Yes. Add an image or video when you want the subject, product, setting, or motion to be grounded in a reference you provide.

How does pricing work?

Pricing is dynamic and usage-based. The amount depends on your selected options and account plan.

smart_toyFor AI products and agents

One agent boundary for video creation

Your product sends a user’s creative request and optional file context to the FLUX Video Agent. The wrapper returns user-facing text and structured video results your interface can render and persist.

  • check_circle One public operation: run_agent
  • check_circle Natural-language requests with optional media context
  • check_circle Reviewable results for follow-up turns and product UI
Login to access the agent
movie

Agent request

Product launch clip

Prompt

Create a 9:16 product story from this image. Add a slow camera push-in, soft studio light, and a subtle ambient soundtrack.

Status

ready to review

Output

video/mp4

Result

User-facing text plus structured video output

Agent contract

Keep the integration simple: pass the user’s request, include file context only when needed, and preserve the structured result for your UI and next turn.

Public operation

run_agent

smart_toy

Runs the FLUX Video Agent for a user’s creative request and returns the user-facing answer plus structured output metadata.

Required input

  • prompt

Optional inputs

  • file_urls[]
  • reference_files[]

Useful result fields

  • response_text
  • tool_result
  • agent_runtime

A result your product can use

The wrapper gives your product a clear user-facing answer alongside structured output, so the interface can show the result without reconstructing state from prose.

response_text

The message your user should read.

tool_result

Structured video result and status.

agent_runtime

Execution context for the agent turn.

video output

A renderable, user-owned MP4 artifact when complete.

{
  "response_text": "Your FLUX video is ready to review.",
  "tool_result": { "status": "succeeded", "artifacts": [{ "kind": "video", "mime_type": "video/mp4" }] },
  "agent_runtime": { "status": "completed" }
}

Quickstart

Pass a creative brief to the agent and keep the structured result available for your interface and the next user turn.

Three steps

  1. 1. Discover run_agent through your agent connection.
  2. 2. Send the request with optional file context.
  3. 3. Render response_text and preserve the structured result.
const result = await agent.callTool("run_agent", {
  prompt: "Create a vertical product video with a slow camera push-in and soft studio light.",
  file_urls: ["https://example.com/product-reference.jpg"]
});

const answer = result.structuredContent?.response_text;
const videoResult = result.structuredContent?.tool_result;

Access

The agent wrapper is designed for account-scoped use so user media, generated artifacts, and follow-up state stay associated with the authenticated Infoseek account.

shield_lock

Authenticated agent access

Use Infoseek authentication when your AI product needs to work with user-owned reference files, generated videos, and private follow-up state.

account_balance_wallet

Dynamic usage-based pricing

Pricing is calculated from the selected options and account plan. Eligibility is checked when the user proceeds.

Built for practical video work

campaign

Social content

Create vertical clips, launch teasers, and short explainers from a simple brief.

edit_video

Focused edits

Make one clear change to an existing take without starting the whole idea over.

shopping_bag

Product storytelling

Animate product images, direct the camera, and create reusable launch assets.

smart_toy

Agent platforms

Add multimodal video creation to an assistant without exposing workflow complexity to your users.