Luma AI connector for AI agents

Tools your video agent reaches for on Luma AI, scoped per user.

CALL ANY TOOL

14 tools for AI video in Luma AI: create, extend, modify, reframe, and upscale videos with Ray, add audio, generate and reframe images with Photon, and list camera concepts, generations, and credits.

lumalabs_add_audio

Add audio

Generate a soundtrack and sound effects for a completed Luma AI video generation from a text prompt. Returns a generation object for the audio job; the video with audio appears in its assets.video once state is completed. Use this to add sound to a video made with Luma AI. The source must be a completed video generation. Poll lumalabs_get_generation with the returned id until state is completed or failed.

Parameters

Name

Type

Required

Description

generation_id

string

Required

ID (UUID) of a Luma AI generation, as returned by a create tool or lumalabs_list_generations. It must be a completed video generation. Example: 123e4567-e89b-12d3-a456-426614174000

prompt

string

Required

Text describing the audio to generate. Example: Crunching snow footsteps and a distant wind howl

negative_prompt

string

Optional

Sounds to avoid in the generated audio. Example: music, speech

lumalabs_create_image

Create image

Generate an image with Luma AI's Photon models from a text prompt, optionally guided by reference images, a style image, a character's photos, or an image to edit. Returns the generation object with its id and state; the image URL appears in assets.image once state is completed (set sync to true to wait for it in the same call). Use this for still images and image edits. Use lumalabs_reframe_image to change an existing image's aspect ratio, and lumalabs_create_video for video.

Parameters

Name

Type

Required

Description

prompt

string

Required

Text describing the image to generate, or the change to make when modify_image_url is set. Example: A teddy bear in sunglasses playing electric guitar

aspect_ratio

string

Optional

Aspect ratio of the output: one of 1:1, 16:9, 9:16, 4:3, 3:4, 21:9 or 9:21. Example: 9:16 Leave empty for the server default (16:9). One of: `1:1`, `16:9`, `9:16`, `4:3`, `3:4`, `21:9`, `9:21`.

character_ref_images

array

Optional

Up to 4 public photo URLs of the same person, used to keep that character consistent in the image. More photos give a better likeness. Example: ["https://example.com/person-1.jpg"]

format

string

Optional

Output file format: jpg or png. Leave empty for jpg. Example: png. One of: `jpg`, `png`.

image_ref

array

Optional

Up to 4 reference images that guide the content of the generation, each with a public url and an optional weight from 0 to 1 that sets its influence. Useful for variations of an image or hard-to-describe concepts. Example: [{"url": "https://example.com/ref.jpg", "weight": 0.85}]

model

string

Optional

Photon image model: photon-1 (higher quality) or photon-flash-1 (faster and cheaper). Defaults to photon-1. Example: photon-flash-1. One of: `photon-1`, `photon-flash-1`. Default: `photon-1`.

modify_image_url

string

Optional

Public URL of an existing image to edit according to the prompt (image-to-image). Example: https://example.com/garden.jpg

modify_image_weight

number

Optional

How closely the result follows modify_image_url, from 0 to 1: higher stays closer to the input, lower is more creative. For color changes Luma recommends 0.0 to 0.1. Only used with modify_image_url. Example: 0.8

style_ref

array

Optional

Style reference images whose look is applied to the generation, each with a public url and an optional weight from 0 to 1. Example: [{"url": "https://example.com/style.jpg", "weight": 0.8}]

sync

boolean

Optional

Set to true to wait for the image to finish and return the completed generation in this call, instead of polling lumalabs_get_generation. Example: true

sync_timeout

number

Optional

Seconds to wait when sync is true (server default 60; capped at 100 by this tool). If the image is not done in time, poll lumalabs_get_generation. Example: 60

lumalabs_create_video

Create video

Start a Luma AI Ray video generation from a text prompt, optionally guided by a start and/or end image and camera concepts. Returns the new generation object with its id and state (queued); the video URL appears in assets.video once state is completed. Use this for text-to-video and image-to-video. Use lumalabs_extend_video to continue or bridge an existing generated video, and lumalabs_create_image for still images. Poll lumalabs_get_generation with the returned id until state is completed or failed.

Parameters

Name

Type

Required

Description

prompt

string

Required

Text describing the video: subject, action, setting and style (3 to 5000 characters). Camera moves can also be written in the prompt, for example 'camera orbit left'. Example: A tiger walking through fresh snow at dawn, cinematic

aspect_ratio

string

Optional

Aspect ratio of the output: one of 1:1, 16:9, 9:16, 4:3, 3:4, 21:9 or 9:21. Example: 9:16 Leave empty for the server default (16:9). One of: `1:1`, `16:9`, `9:16`, `4:3`, `3:4`, `21:9`, `9:21`.

concepts

array

Optional

Camera concept keys that steer camera movement and framing, combined in one shot. Get valid keys from lumalabs_list_concepts; common ones are dolly_zoom, orbit_left, push_in, aerial_drone and handheld. Example: ["dolly_zoom", "handheld"]

duration

string

Optional

Video length: 5s or 9s. Leave empty for the default (5s). Example: 5s. One of: `5s`, `9s`.

end_image_url

string

Optional

Public URL of an image to use as the last frame; the video animates toward it. Can be combined with start_image_url to animate between two images. Example: https://example.com/end-frame.jpg

loop

boolean

Optional

Set to true to make the video loop seamlessly (last frame flows into the first). Not supported when both a start and an end image are given. Example: true

model

string

Optional

Ray video model to use: ray-2 (higher quality) or ray-flash-2 (faster and cheaper). Defaults to ray-2. Example: ray-flash-2. One of: `ray-2`, `ray-flash-2`. Default: `ray-2`.

resolution

string

Optional

Output resolution: 540p, 720p, 1080p or 4k. Higher resolutions cost more. Leave empty for the model default. Example: 720p. One of: `540p`, `720p`, `1080p`, `4k`.

start_image_url

string

Optional

Public URL of an image to use as the first frame (image-to-video). Must be reachable by Luma's servers; upload it to your own storage or CDN first. Example: https://example.com/start-frame.jpg

lumalabs_delete_generation

Delete generation

lumalabs_extend_video

Extend video

Start a Luma AI Ray video generation that continues an existing generated video forward, leads up to it backward, or bridges it to an image or another generated video. Returns the new generation object with its id and state (queued); the extended clip appears in assets.video once state is completed. Use this only for videos made with Luma AI. Use lumalabs_create_video for a new video from text or images. The source generation must be in state completed; check it with lumalabs_get_generation.

Parameters

Name

Type

Required

Description

generation_id

string

Required

ID (UUID) of a Luma AI generation, as returned by a create tool or lumalabs_list_generations. It must be a completed video generation. Example: 123e4567-e89b-12d3-a456-426614174000

prompt

string

Required

Text describing what happens in the new footage: subject, action, setting and style (3 to 5000 characters). Camera moves can also be written in the prompt, for example 'camera orbit left'. Example: A tiger walking through fresh snow at dawn, cinematic

aspect_ratio

string

Optional

Aspect ratio of the output: one of 1:1, 16:9, 9:16, 4:3, 3:4, 21:9 or 9:21. Match the source video's aspect ratio. Example: 9:16 Leave empty for the server default (16:9). One of: `1:1`, `16:9`, `9:16`, `4:3`, `3:4`, `21:9`, `9:21`.

concepts

array

Optional

direction

string

Optional

forward continues the source video after its last frame; backward generates footage that leads up to its first frame. Defaults to forward. Example: backward. One of: `forward`, `backward`. Default: `forward`.

duration

string

Optional

Video length: 5s or 9s. Leave empty for the default (5s). Example: 5s. One of: `5s`, `9s`.

loop

boolean

Optional

Set to true to make the result loop. Only supported when extending forward with no other end set. Example: false

model

string

Optional

Ray video model to use: ray-2 (higher quality) or ray-flash-2 (faster and cheaper). Defaults to ray-2. Example: ray-flash-2. One of: `ray-2`, `ray-flash-2`. Default: `ray-2`.

other_end_generation_id

string

Optional

Optional ID of a second completed Luma AI video generation to interpolate to, producing a clip that bridges the two videos. Do not combine with other_end_image_url. Example: 9b2f6c1e-4d3a-4f8b-a1c2-7e5d3b9a0f14

other_end_image_url

string

Optional

Optional public image URL for the opposite end of the new clip: the final frame when extending forward, the opening frame when extending backward. Do not combine with other_end_generation_id. Example: https://example.com/final-frame.jpg

resolution

string

Optional

Output resolution: 540p, 720p, 1080p or 4k. Higher resolutions cost more. Leave empty for the model default. Example: 720p. One of: `540p`, `720p`, `1080p`, `4k`.

lumalabs_get_credits

Get credits

lumalabs_get_generation

Get generation

lumalabs_list_camera_motions

List camera motions

lumalabs_list_concepts

List concepts

lumalabs_list_generations

List generations

lumalabs_modify_video

Modify video

lumalabs_reframe_image

Reframe image

Change an image's aspect ratio with Luma AI Photon by generating new content around it (outpainting). Returns the new generation object with its id and state (queued); the reframed image appears in assets.image once state is completed. Use this to adapt an existing image to another aspect ratio. Use lumalabs_create_image to generate or edit an image, and lumalabs_reframe_video for video. Poll lumalabs_get_generation with the returned id until state is completed or failed.

Parameters

Name

Type

Required

Description

aspect_ratio

string

Required

Aspect ratio of the output: one of 1:1, 16:9, 9:16, 4:3, 3:4, 21:9 or 9:21. This is the target aspect ratio; Luma generates content for the new area. Example: 9:16. One of: `1:1`, `16:9`, `9:16`, `4:3`, `3:4`, `21:9`, `9:21`.

media_url

string

Required

Public URL of the source image to reframe (up to 10 MB). Example: https://example.com/portrait.jpg

model

string

Required

Photon image model used to reframe: photon-1 (higher quality) or photon-flash-1 (faster and cheaper). Example: photon-1. One of: `photon-1`, `photon-flash-1`.

format

string

Optional

Output file format: jpg or png. Leave empty for jpg. Example: png. One of: `jpg`, `png`.

grid_position_x

integer

Optional

Advanced, optional. Horizontal pixel offset at which to place the source inside the new canvas. All positioning values are pixels on the target canvas for the chosen aspect ratio (for video, 16:9 is 1280x720; for images, 16:9 is 2048x1152). Leave empty to let Luma fit the source automatically. Example: 120

grid_position_y

integer

Optional

Advanced, optional. Vertical pixel offset at which to place the source inside the new canvas. All positioning values are pixels on the target canvas for the chosen aspect ratio (for video, 16:9 is 1280x720; for images, 16:9 is 2048x1152). Leave empty to let Luma fit the source automatically. Example: 0

prompt

string

Optional

Optional text steering what Luma generates in the newly added area. Example: Continue the city skyline to the left and right

resized_height

integer

Optional

Advanced, optional. Height, in pixels, to resize the source to before placing it. All positioning values are pixels on the target canvas for the chosen aspect ratio (for video, 16:9 is 1280x720; for images, 16:9 is 2048x1152). Leave empty to let Luma fit the source automatically. Example: 720

resized_width

integer

Optional

Advanced, optional. Width, in pixels, to resize the source to before placing it. All positioning values are pixels on the target canvas for the chosen aspect ratio (for video, 16:9 is 1280x720; for images, 16:9 is 2048x1152). Leave empty to let Luma fit the source automatically. Example: 1280

x_end

integer

Optional

Advanced, optional. Right edge, in pixels, of the crop bounds applied to the source. All positioning values are pixels on the target canvas for the chosen aspect ratio (for video, 16:9 is 1280x720; for images, 16:9 is 2048x1152). Leave empty to let Luma fit the source automatically. Example: 1280

x_start

integer

Optional

Advanced, optional. Left edge, in pixels, of the crop bounds applied to the source. All positioning values are pixels on the target canvas for the chosen aspect ratio (for video, 16:9 is 1280x720; for images, 16:9 is 2048x1152). Leave empty to let Luma fit the source automatically. Example: 0

y_end

integer

Optional

Advanced, optional. Bottom edge, in pixels, of the crop bounds applied to the source. All positioning values are pixels on the target canvas for the chosen aspect ratio (for video, 16:9 is 1280x720; for images, 16:9 is 2048x1152). Leave empty to let Luma fit the source automatically. Example: 720

y_start

integer

Optional

Advanced, optional. Top edge, in pixels, of the crop bounds applied to the source. All positioning values are pixels on the target canvas for the chosen aspect ratio (for video, 16:9 is 1280x720; for images, 16:9 is 2048x1152). Leave empty to let Luma fit the source automatically. Example: 0

lumalabs_reframe_video

Reframe video

Change a video's aspect ratio with Luma AI Ray by generating new content around it (for example landscape to vertical). Returns the new generation object with its id and state (queued); the reframed video appears in assets.video once state is completed. Use this to adapt a video to another aspect ratio. Use lumalabs_reframe_image for still images and lumalabs_modify_video to restyle a video. Poll lumalabs_get_generation with the returned id until state is completed or failed.

Parameters

Name

Type

Required

Description

aspect_ratio

string

Required

media_url

string

Required

Public URL of the source video to reframe (up to 100 MB; up to 10s for ray-2 or 30s for ray-flash-2). Example: https://example.com/landscape.mp4

model

string

Required

Ray video model used to reframe the video: ray-2 (higher quality) or ray-flash-2 (faster and cheaper; accepts longer source videos). Example: ray-2. One of: `ray-2`, `ray-flash-2`.

first_frame_url

string

Optional

Optional public URL of an image showing the desired first frame of the reframed video. Example: https://example.com/reframed-first-frame.jpg

grid_position_x

integer

Optional

grid_position_y

integer

Optional

prompt

string

Optional

Optional text steering what Luma generates in the newly added area. Example: Extend the beach and sky on both sides

resized_height

integer

Optional

resized_width

integer

Optional

x_end

integer

Optional

x_start

integer

Optional

y_end

integer

Optional

y_start

integer

Optional

lumalabs_upscale_video

Upscale video

Build your Agent

Same auth pattern across LangChain, OpenAI, Anthropic, and Google ADK.

import { ScalekitClient } from "@scalekit-sdk/node";
import { createReactAgent } from "@langchain/langgraph/prebuilt";

const sk = new ScalekitClient(env.SCALEKIT_ENV_URL, env.SCALEKIT_CLIENT_ID, env.SCALEKIT_CLIENT_SECRET);

// Luma AI tools scoped to this user
const { tools } = await sk.tools.listScopedTools("user_123", {
  filter: { connectionNames: ["lumalabs"], toolNames: [
    "lumalabs_create_video",
    "lumalabs_get_generation",
    "lumalabs_upscale_video"] },
  pageSize: 100,
});

const agent = createReactAgent({ llm, tools });
await agent.invoke({ messages: [{ role: "user", content: "Create a drone video of a coastline at sunset and upscale it to 4k" }] });

import OpenAI from "openai";
import { ScalekitClient } from "@scalekit-sdk/node";

const sk = new ScalekitClient(env.SCALEKIT_ENV_URL, env.SCALEKIT_CLIENT_ID, env.SCALEKIT_CLIENT_SECRET);
const openai = new OpenAI();

const { tools } = await sk.tools.listScopedTools("user_123", {
  filter: { connectionNames: ["lumalabs"] }, pageSize: 100,
});

const res = await openai.chat.completions.create({
  model: "gpt-5",
  messages: [{ role: "user", content: "Create a drone video of a coastline at sunset and upscale it to 4k" }],
  tools,
});

// Execute the tool call with the user's vaulted Luma AI access token
await sk.tools.executeTool(res.choices[0].message.tool_calls[0], "user_123");

import Anthropic from "@anthropic-ai/sdk";
import { ScalekitClient } from "@scalekit-sdk/node";

const sk = new ScalekitClient(env.SCALEKIT_ENV_URL, env.SCALEKIT_CLIENT_ID, env.SCALEKIT_CLIENT_SECRET);
const anthropic = new Anthropic();

const { tools } = await sk.tools.listScopedTools("user_123", {
  filter: { connectionNames: ["lumalabs"] }, pageSize: 100,
});

const msg = await anthropic.messages.create({
  model: "claude-sonnet-5",
  max_tokens: 1024,
  messages: [{ role: "user", content: "Create a drone video of a coastline at sunset and upscale it to 4k" }],
  tools,
});

// Tool call runs with the user's vaulted Luma AI access token
await sk.tools.executeTool(msg.content, "user_123");

import { Agent } from "@google/adk/agents";
import { ScalekitClient } from "@scalekit-sdk/node";

const sk = new ScalekitClient(env.SCALEKIT_ENV_URL, env.SCALEKIT_CLIENT_ID, env.SCALEKIT_CLIENT_SECRET);

const { tools } = await sk.tools.listScopedTools("user_123", {
  filter: { connectionNames: ["lumalabs"] }, pageSize: 100,
});

const agent = new Agent({
  name: "video_agent",
  model: "gemini-2.5-pro",
  instruction: "Luma AI tools scoped to this user",
  tools,
});

await agent.run("Create a drone video of a coastline at sunset and upscale it to 4k");

Try these prompts

Copy any prompt into your agent. Each maps directly to a Luma AI tool. Click to copy, paste into your agent, done.

Video

Copy the prompt

Copied

Create a 16:9 video of a drone flying over a coastline at sunset.

Copy the prompt

Copied

Extend the coastline video forward past the lighthouse.

Copy the prompt

Copied

Upscale the finished coastline video to 4k.

Edit video

Copy the prompt

Copied

Reframe this landscape video to vertical for Reels.

Copy the prompt

Copied

Restyle this clip as a watercolor animation.

Copy the prompt

Copied

Add ocean sounds and a soft piano soundtrack to the video.

Images and account

Copy the prompt

Copied

Generate a product image of a ceramic vase with Photon.

Copy the prompt

Copied

Reframe this product image to 9:16.

Copy the prompt

Copied

How many credits are left on our Luma AI account?

SEE HOW AUTH WORKS

Each user connects their own Luma AI access token once, and Scalekit sends it with every call. Access tokens stay vaulted, every call is scope checked, and every action is logged.

Authorize

Your user connects

Luma AI

once. We tie it to their identity and the meetings they approved — no shared bot account, no org-wide access

Who:

user ‘A’

when:

Once per user

access:

Limited to user

Store

Their

Luma AI

token lives in a vault scoped to them. User A's meetings are never reachable by an agent acting for user B, even on the same connection

vault:

encrypted

scope:

per-user

tokens:

auto-refreshed

Resolve

When your agent calls a

Luma AI

tool, we fetch the right token server-side. It never touches your agent, never appears in the LLM context, never shows up in your logs

speed:

~40ms

check:

before every call

seen by:

nobody

Audit

Every

Luma AI

tool call is logged — who triggered it, which meeting was fetched, what came back. 90 days of history, tied to the user who authorized it

history:

90 days

export:

SIEM-ready

logged:

every call

Test other agents

See the same per-user auth pattern across other AI and media connectors.

GTM and RevOps Teams

Competitive intelligence briefing agent

Scans Gong calls for competitor mentions, matches each one to its Notion battlecard, and DMs every affected rep a single Slack digest per cycle. Every call runs as the PMM who owns the briefing, never a shared bot.

Why Scalekit

Secure your agent's access. Connectors ship in minutes

01.

Shared tokens break per-user analytics

A shared Luma AI access token looks fine in a demo. In production every video generation looks like one service account, and you cannot tell which user triggered it. Scalekit resolves the credential of the actual user who triggered the agent, never a shared bot.

// shared key
audit → bot_service_account

// scalekit
audit → user_abc ✓

02.

Authentication is not authorization

03.

Multi-tenancy is architectural

04.

Luma AI today. Ten connectors tomorrow.

“Our agents act across Salesforce, Gong, Google Drive, and more, on behalf of every customer. Scalekit behind the scenes meant we can keep adding tools without ever rebuilding how credentials or tool calling work.”

Venu Madhav Kattagoni

Head of Engineering / Von

FAQs

Frequently Asked Questions

Does the agent access Luma AI as the user or through a shared access token?

As the user. Each user connects their own Luma AI access token once, and Scalekit sends it with every call. Audit logs attribute every video generation to that user, not a shared service account.

Where is the Luma AI access token stored?

In Scalekit's managed AES-256 token vault, namespaced per tenant. To rotate it, the user reconnects with a new access token. Revocation is a single dashboard action. Credentials never appear in prompts, logs, or LLM context.

Can I limit what the agent does in Luma AI?

Yes. Filter by tool name in listScopedTools to expose only what you want, for example video and image generation without lumalabs_delete_generation, which permanently deletes a generation. Scalekit also enforces scope checks before every API call.

What happens when a user revokes Luma AI access?

The connection is invalidated on the next tool call. Subsequent requests for that user fail closed with a clear error. Other users in the tenant remain unaffected. The event is logged for audit.

Which camera moves can the agent ask for?

lumalabs_list_concepts returns the camera concept keys Ray accepts, such as dolly zoom, orbit left, push in, and aerial drone, and lumalabs_list_camera_motions lists the camera motion phrases Luma AI recognizes in video prompts.