@everyfeed/sdk

Type-safe ESM client for the EveryFeed Public API. Node.js 22 or newer is required.

npm install @everyfeed/sdk

import { EveryFeedClient } from "@everyfeed/sdk";

const everyfeed = new EveryFeedClient({
  baseUrl: "https://api.everyfeed.app",
  token: process.env.EVERYFEED_API_TOKEN!,
});

const connection = await everyfeed.connections.start("linkedin", {
  redirectUrl: "https://app.everyfeed.app/app/settings",
  timezone: "Europe/Istanbul",
});

console.log(connection.data.url);

The token option accepts either an organization API token with the evf_ prefix or a Public API-audience OAuth access token with the evfo_at_ prefix. OAuth refresh tokens, authorization codes, and client secrets are not accepted.

Custom credential providers use a retry-safe second step:

const custom = await everyfeed.connections.connectCustom(
  "wordpress",
  {
    credentials: {
      password: process.env.WORDPRESS_APPLICATION_PASSWORD!,
      url: "https://blog.example.com",
      username: "everyfeed",
    },
  },
  { idempotencyKey: crypto.randomUUID() },
);

For OAuth providers that require an account or page choice after the callback:

const connectionId = "PENDING_CONNECTION_UUID";
const pending = await everyfeed.connections.selectionOptions(connectionId);
const refreshed = await everyfeed.connections.refreshSelectionOptions(
  connectionId,
  { idempotencyKey: crypto.randomUUID() },
);
await everyfeed.connections.selectOption(
  connectionId,
  { selectionId: refreshed.data.selectionOptions![0]!.id },
  { idempotencyKey: crypto.randomUUID() },
);

Writes that can create, update, or delete resources accept an idempotency key:

const image = await everyfeed.media.generateImage(
  {
    enhancePrompt: true,
    prompt: "A clean product launch illustration",
    style: "Minimalist",
  },
  { idempotencyKey: crypto.randomUUID() },
);

AI images use the Postiz-compatible fixed square pipeline. Set enhancePrompt: true to expand the prompt with GPT-4.1. Selecting any SDK EveryFeedAiImageStyle also enables enhancement before chatgpt-image-latest generates the image.

AI video creation accepts up to three image references. Read everyfeed.media.listVideoProviders() first when building a dynamic client:

await everyfeed.media.createVideo(
  {
    prompt: "A short vertical product reveal",
    referenceImages: [
      { file: campaignHeroBlob, fileName: "campaign-hero.png" },
      { file: productBlob, fileName: "product.png" },
    ],
    seconds: 8,
  },
  { idempotencyKey: crypto.randomUUID() },
);

See the repository's docs/public-api.md for endpoints, scopes, pagination, media upload, and asynchronous video-generation examples.

License

MIT