Generated banner for Zoom Apps SDK
Generated logo for Zoom Apps SDK

Zoom Apps SDK

OpenAIOfficialCreated by ZoomDistributed by OpenAI

  • Skill
  • Tool integration
  • Software development
  • MIT
  • Medium risk

From the author

Use when using Apps SDK.
— the description by Zoom

AutoGineer summary

Written by AutoGineer from the skill’s own files, not by the author

The Zoom Apps SDK skill provides a structured approach for developers building applications that run directly inside the Zoom client. It covers the lifecycle of a Zoom App, from initial configuration and context verification to implementing advanced features like the Layers API, in-client OAuth, and inter-instance communication. The skill emphasizes the importance of identifying the correct running context—such as meetings, webinars, or the main client—to ensure features are gated appropriately and the user experience remains consistent across different Zoom environments.

What it does

  • Initializes the Zoom Apps SDK with required capabilities.
  • Identifies the current running context, such as meeting or main client.
  • Facilitates in-client OAuth flows for secure user authentication.
  • Enables communication between different app instances.
  • Provides access to meeting-specific data and participant information.
  • Supports advanced UI and media features like virtual backgrounds and notification management.

When to use it

  • When building an application intended to run inside the Zoom client interface.
  • When you need to access meeting context, participant lists, or recording status.
  • When implementing custom UI overlays or media controls within a Zoom meeting.
  • When you need to authenticate users without redirecting them to an external browser.

Advantages

  • Provides a unified API for interacting with various Zoom client contexts.
  • Supports secure, in-client authentication workflows.
  • Enables deep integration with Zoom meeting features and participant data.
  • Offers clear patterns for handling cross-instance communication.

Usage examples

“How do I initialize the Zoom Apps SDK?”
You must call zoomSdk.config() first, passing in the required capabilities and version to initialize the SDK and declare the features your app intends to use.
“How can I get the current meeting ID?”
After initializing the SDK, you can call zoomSdk.getMeetingContext() to retrieve the meetingID, meetingTopic, and meetingStatus.
“How do I trigger authentication within the Zoom client?”
Use the zoomSdk.authorize() method to trigger the in-client OAuth flow, which avoids external browser redirects and fires an onAuthorized event upon completion.
“How do I send a message to another instance of my app?”
First call zoomSdk.connect() to establish the connection, then use zoomSdk.postMessage() to send a payload to other app instances.

Install

  1. Download the skill as a ZIP:

    https://autogineer.com/agent-skills/openai/zoom-apps-sdk/download
  2. Unzip it into the cross-client skills folder of your project:

    mkdir -p .agents/skills && unzip zoom-apps-sdk.zip -d .agents/skills

Documentation by Zoom

Zoom Apps SDK

Use this skill when the app runs inside the Zoom client. If the user only needs to embed a meeting in an external app, route to Meeting SDK instead.

Workflow

  1. Confirm the running context: meeting, webinar, main client, phone, collaborate mode, immersive mode, camera mode, or Layers API.

  2. Configure Marketplace app settings, allowed domains, redirect URIs, and in-client OAuth before implementing SDK calls.

  3. Initialize zoomSdk and gate features by capability and running context.

  4. Design client communication and data flow: frontend SDK calls, backend REST calls, in-client OAuth tokens, and webhook handoff.

  5. Implement advanced client features only after the base app loads reliably: Layers API, breakout rooms, guest mode, collaborate mode, or ZMail.

  6. Debug blank panels, domain allowlist issues, CSP, missing capabilities, running-context mismatch, and OAuth redirect problems independently.

References

Files

references/layers-api.md

Zoom Apps SDK - Layers API Reference

Build immersive video layouts and camera overlays using the Layers API.

Overview

The Layers API (v1.5) provides rendering modes for custom visual experiences. Requires Zoom Client v5.10.6+.

ModeDescriptionUse Case
Team (immersive + person cutout)Canvas with background-removed participant cutoutsPodcast, talk show, classroom
Presentation (immersive + rectangle cutout)Canvas with full-width participant video tilesPresentations, branded meetings
CameraOverlay on user's own camera feed (OSR)Branding, name tags, effects
ControllerSidebar app that coordinates Layers modesRequired for all modes above

Note: When using the Layers API, your app is categorized as an "Immersive App" on the Marketplace.

Required Capabilities

await zoomSdk.config({
  capabilities: [
    'getRunningContext',
    'runRenderingContext', 'closeRenderingContext',
    'drawParticipant', 'clearParticipant',
    'drawImage', 'clearImage',
    'drawWebView', 'clearWebView',
    'postMessage', 'onMessage',
    'sendAppInvitationToAllParticipants',
    'onMyMediaChange',
    'onRenderedAppOpened'
  ],
  version: '0.16'
});

Gotcha: The official guide lists clearWebview (lowercase 'v') in one config example. Use clearWebView (camelCase) to match the actual method name.

Types

PixelValue

All position/size parameters accept three formats:

type PixelValue = `${string}px` | `${string}%` | number;
FormatExampleMeaning
"Npx""100px"CSS reference pixels
"N%""50%"Percentage of container/view
number1280Raw physical pixels
ParticipantCutoutShape
type ParticipantCutoutShape =
  | "person"             // v5.9.3+  — Cut out background (AI segmentation)
  | "standard"           // v5.11.3+ — Full uncropped video (squared corners)
  | "rectangle"          // v5.11.0+ — Rounded rectangle (30px radius)
  | "circle"             // v5.11.3+ — Circle
  | "square"             // v5.11.3+ — Square (30px radius)
  | "verticalRectangle"  // v5.11.3+ — Vertical rectangle (30px radius)

All shapes have 30px rounded corners except "standard" which has squared corners.

RenderingContextView
type RenderingContextView = "immersive" | "camera";

Lifecycle

Starting a Rendering Context
// Team mode (person cutout — removes backgrounds)
await zoomSdk.runRenderingContext({
  view: 'immersive',
  defaultCutout: 'person'
});

// Presentation mode (rectangle cutout — keeps backgrounds)
await zoomSdk.runRenderingContext({
  view: 'immersive',
  defaultCutout: 'rectangle'
});

// Camera mode (affects only your video stream)
await zoomSdk.runRenderingContext({ view: 'camera' });

runRenderingContext(options):

  • view (required): "immersive" | "camera"

  • defaultCutout (optional): Sets the default cutout shape for all drawParticipant() calls in this context

Running Context Values
ContextMeaning
inMeetingDefault sidebar panel
inImmersiveRunning in immersive mode (team or presentation)
inCameraRunning as virtual camera (off-screen rendering)
const { runningContext } = await zoomSdk.getRunningContext();
// runningContext changes automatically when runRenderingContext() is called
Updating Content

To move, resize, or adjust a drawn element: clear it first, then redraw.

// Move a participant
await zoomSdk.clearParticipant({ participantUUID: uuid });
await zoomSdk.drawParticipant({ participantUUID: uuid, x: 100, y: 200, width: 640, height: 480, zIndex: 1 });

There is no in-place update — always clear + redraw.

Closing
await zoomSdk.closeRenderingContext();
// Returns app to sidebar, runningContext becomes "inMeeting"
Constraints
  • Only a meeting host can set the rendering context to immersive

  • Only one immersive context can exist at a time (second attempt fails with error)

  • Camera mode + Presentation mode can run simultaneously

  • Host must use sendAppInvitationToAllParticipants to transition other participants

  • If aomhost package needs download, runRenderingContext returns non-success

Drawing Methods

drawParticipant

Position a participant's video feed on the canvas.

drawParticipant(options: DrawParticipantOptions): Promise<GeneralMessageResponse>
ParameterTypeDefaultDescription
participantUUIDstring—Meeting-specific participant identifier
participantIdstring—DEPRECATED — use participantUUID
xPixelValue"0px"Horizontal position
yPixelValue"0px"Vertical position
widthPixelValue"100%"Width (aspect ratio maintained)
heightPixelValue"100%"Height (aspect ratio maintained)
zIndexnumber1Stacking order (higher = on top)
cutoutParticipantCutoutShapecontext defaultCutout behavior (v5.9.3+)
cameraModeMirroringbooleanfalseMirror video in camera mode (v5.13.5+)

Mode differences:

  • Immersive: Can draw any participant

  • Camera: Can only draw current user (self)

// Immersive — draw any participant with person cutout
await zoomSdk.drawParticipant({
  participantUUID: 'uuid-from-getMeetingParticipants',
  x: 40, y: 100,
  width: 580, height: 500,
  zIndex: 1,
  cutout: 'person'
});

// Camera — draw self with mirroring
await zoomSdk.drawParticipant({
  participantUUID: myUUID,
  x: 0, y: 0,
  width: 1280, height: 720,
  zIndex: 1,
  cameraModeMirroring: true  // v5.13.5+
});
drawImage

Draw static images (backgrounds, overlays, borders).

drawImage(options: DrawImageOptions): Promise<DrawImageResponse>
ParameterTypeDefaultDescription
imageDataImageData—Required. Standard JS ImageData object (width, height, pixel bytes)
xPixelValue"0px"Horizontal position
yPixelValue"0px"Vertical position
zIndexnumber1Stacking order

Returns: { imageId: string } — use this ID with clearImage().

Important: imageData is a standard JavaScript ImageData object (from canvas.getImageData()), NOT a base64 data URL.

const canvas = document.createElement('canvas');
canvas.width = 1280;
canvas.height = 720;
const ctx = canvas.getContext('2d');
// ... draw on canvas ...
const imageData = ctx.getImageData(0, 0, canvas.width, canvas.height);

const { imageId } = await zoomSdk.drawImage({
  imageData,
  x: 0, y: 0,
  zIndex: 0
});
HiDPI Constraints

drawImage() does not directly support HiDPI image sizes. For HiDPI/Retina:

  1. Draw to canvas using the scaling ratio (window.devicePixelRatio)

  2. Divide out the ratio for x/y coordinates when passing to drawImage

  3. Keep the ratio for width and height

  4. You may need to tile the screen for full-screen images

const dpr = window.devicePixelRatio || 1;
const canvas = document.createElement('canvas');
canvas.width = 1280 * dpr;
canvas.height = 720 * dpr;
const ctx = canvas.getContext('2d');
ctx.scale(dpr, dpr);
// ... draw at logical pixels ...
const imageData = ctx.getImageData(0, 0, canvas.width, canvas.height);

await zoomSdk.drawImage({
  imageData,
  x: 0, y: 0,
  zIndex: 0
});
drawWebView

Position the app's OSR (Off-Screen Rendering) webview within the Layers canvas.

drawWebView(options: DrawWebViewOptions): Promise<GeneralMessageResponse>
ParameterTypeDefaultDescription
xPixelValue0Horizontal position in OSR target area
yPixelValue0Vertical position in OSR target area
widthPixelValuefull rendering widthWidth in OSR target area
heightPixelValuefull rendering heightHeight in OSR target area
zIndexnumber1Stacking order

⚠ Documentation inconsistency: The official Zoom guides show a webviewId parameter in examples, but the TypeDoc type definition (v0.16.36) does not include it. Since there is only one webview per app, this parameter may be vestigial. If in doubt, omit it.

What the webview renders: Your app's home URL as configured in zoomSdk.config(). It's an off-screen rendering of your app — not a configurable URL.

Only one webview per rendering context. There is no multi-webview support.

// Full-screen webview in camera mode
const config = await zoomSdk.config({ /* ... */ });
await zoomSdk.runRenderingContext({ view: 'camera' });

await zoomSdk.drawWebView({
  x: 0,
  y: 0,
  width: config.media.renderTarget.width,   // Default: 1280
  height: config.media.renderTarget.height,  // Default: 720
  zIndex: 2
});
// Partial webview overlay (bottom third of camera)
await zoomSdk.drawWebView({
  x: 0,
  y: 480,
  width: 1280,
  height: 240,
  zIndex: 2
});
Webview Communication

The sidebar app and the camera/immersive app are separate instances. Use postMessage() and onMessage to communicate between them:

// Sidebar instance → Camera instance (no connect() required)
zoomSdk.postMessage({ command: 'update-overlay', text: 'Q&A Time' });

// Camera instance listens
zoomSdk.addEventListener('onMessage', (eventInfo) => {
  if (eventInfo.command === 'update-overlay') {
    document.getElementById('overlay-text').textContent = eventInfo.text;
  }
});

Note: connect() is NOT required for app-to-app messaging in Layers. postMessage works between instances of the same app.

Clearing
// Clear participant (use participantUUID, not the deprecated participantId)
await zoomSdk.clearParticipant({ participantUUID: 'uuid' });

// Clear image (use imageId from drawImage response)
await zoomSdk.clearImage({ imageId: 'id-from-drawImage' });

// Clear webview (hides it — app continues running)
await zoomSdk.clearWebView();
// Note: TypeDoc v0.16.36 shows no parameters.
// Guide examples show { webviewId: "xxx" } but this may be outdated.

Coordinate System

  • Origin: Top-left corner (0, 0)

  • X: Increases rightward

  • Y: Increases downward

  • Units: PixelValue — supports "Npx", "N%", or raw number

Immersive Mode
  • Coordinates are CSS pixels relative to the meeting canvas

  • Automatic scaling for different window sizes

Camera Mode
  • Coordinates are raw pixels relative to renderTarget dimensions

  • Default renderTarget: 1280×720 (configurable)

  • Access via: config.media.renderTarget.width / .height

const config = await zoomSdk.config({ /* ... */ });
const rtWidth = config.media.renderTarget.width;   // e.g. 1280
const rtHeight = config.media.renderTarget.height;  // e.g. 720

Z-Index Layering

zIndex: 2+  ─  WebViews, interactive overlays (top)
zIndex: 1   ─  Participant videos
zIndex: 0   ─  Background images (bottom)

Higher zIndex values render on top. All three element types (participant, image, webview) share the same z-index space and can overlap.

Events

onRenderedAppOpened

Fires when the rendering context is ready. Best signal that CEF is initialized in camera mode.

zoomSdk.addEventListener('onRenderedAppOpened', () => {
  // Safe to call drawParticipant, drawImage, drawWebView
});
onMyMediaChange

Fires when the user's video changes (camera switch, "Original ratio" toggle, "HD" toggle). Returns device pixel dimensions of the source video.

zoomSdk.addEventListener('onMyMediaChange', (event) => {
  // event.media.video.width / height — device pixels of source video
  // Redraw your layout if needed
});
Window Resize (Immersive Only)

When the Zoom meeting window is resized, the app must move and resize participants/images. Not relevant to Camera Mode (fixed renderTarget).

Immersive Mode vs Camera Mode

AspectImmersiveCamera
ScopeEntire meeting viewUser's camera only
drawParticipantAny participantSelf only
drawImageYesYes
drawWebViewYesYes
Who sees itAll participantsAll see it on this user's feed
Browser engineStandard WebViewCEF (Chromium Embedded Framework)
RenderingOn-screenOff-screen (OSR)
Coordinate spaceCSS pixelsRaw pixels (renderTarget)
SimultaneousOne immersive at a timeCan run with Presentation mode

Camera Mode: CEF Race Condition

Camera mode uses CEF which takes time to initialize. Draw calls may fail if called too early.

Best approach: Listen for onRenderedAppOpened:

zoomSdk.addEventListener('onRenderedAppOpened', async () => {
  await zoomSdk.drawWebView({ x: 0, y: 0, width: 1280, height: 720, zIndex: 2 });
});

Fallback: Retry with exponential backoff:

async function drawWithRetry(drawFn, maxRetries = 5) {
  for (let i = 0; i < maxRetries; i++) {
    try {
      await drawFn();
      return;
    } catch (error) {
      if (i === maxRetries - 1) throw error;
      await new Promise(r => setTimeout(r, 200 * Math.pow(2, i)));
    }
  }
}

Alternative: Check running context:

const { runningContext } = await zoomSdk.getRunningContext();
if (runningContext === 'inCamera') {
  // CEF is ready, safe to draw
}

Performance Tips

  • Use requestAnimationFrame for animations

  • Minimize draw calls (batch updates when possible)

  • Pre-render complex backgrounds to a single canvas ImageData

  • Keep zIndex values low (0-10 range)

  • Clear unused elements to free resources

  • Test on lower-end hardware

  • For full-screen images: tile the screen (HiDPI limitation)

Example: Two-Person Podcast Layout

// Background
const canvas = document.createElement('canvas');
canvas.width = 1280;
canvas.height = 720;
const ctx = canvas.getContext('2d');
const gradient = ctx.createLinearGradient(0, 0, 1280, 720);
gradient.addColorStop(0, '#1a1a2e');
gradient.addColorStop(1, '#16213e');
ctx.fillStyle = gradient;
ctx.fillRect(0, 0, 1280, 720);
const imageData = ctx.getImageData(0, 0, 1280, 720);

await zoomSdk.drawImage({ imageData, x: 0, y: 0, zIndex: 0 });

// Host (left) — person cutout removes background
await zoomSdk.drawParticipant({
  participantUUID: hostUUID,
  x: 40, y: 100, width: 580, height: 500,
  zIndex: 1, cutout: 'person'
});

// Guest (right)
await zoomSdk.drawParticipant({
  participantUUID: guestUUID,
  x: 660, y: 100, width: 580, height: 500,
  zIndex: 1, cutout: 'person'
});

Version History

FeatureClient VersionSDK Version
Core Layers API5.9.00.16
cutout: "person"5.9.30.16
cutout: "rectangle"5.11.00.16
cutout: "circle", "square", "verticalRectangle"5.11.30.16
drawWebView() / clearWebView()5.10.60.16.11+
Camera Mode5.13.10.16
cameraModeMirroring5.13.50.16

Resources

Requirements and compatibility

Written for: OpenAI Codex

Works with: Claude Code, Claude apps, OpenAI Codex, GitHub Copilot, Cursor, Gemini CLI, Kiro, OpenCode, Goose, Amp, Roo Code, JetBrains Junie, Factory Droid, Devin (ex-Windsurf), Mistral Vibe, OpenHands, Letta

“Works with” means the tool reads the open Agent Skills format; a skill written for one tool may use features others lack.

Security

What our scanner noticed· Medium risk

  • Possible data-exfiltration endpoints

A static scan is not a guarantee. Skills are third-party content: review a skill before you let an agent use it.

Attribution and licence

Created by
Zoom
Publisher
OpenAI (Official)
Distributed by
OpenAI
Source
openai/plugins · plugins/zoom/skills/zoom-apps-sdk · commit d420d13
Licence
MIT License (redistribution allowed)
  • LICENSE (added by AutoGineer from plugins/zoom/LICENSE)
  • ATTRIBUTION.md (written by AutoGineer)
Synced

AutoGineer is not affiliated with or endorsed by Zoom or OpenAI.

Download ZIP

The ZIP holds the skill’s files, its licence files and an ATTRIBUTION.md. Download zoom-apps-sdk.zip