@latimer-woods-tech/video

Cloudflare Stream and R2 client wrappers for the Factory automated video production pipeline.

Overview

@latimer-woods-tech/video provides type-safe, platform-neutral wrappers around:

• Cloudflare Stream REST API — upload, retrieve, list, and delete videos
• Cloudflare R2 bucket bindings — store/fetch raw video files and narration audio

It is used by the @latimer-woods-tech/schedule package and the apps/video-studio render pipeline.

Installation

npm install @latimer-woods-tech/video

Environment bindings

interface VideoEnv {
  CF_ACCOUNT_ID: string;   // Cloudflare account ID
  CF_STREAM_TOKEN: string; // API token with Stream:Edit + Stream:Read
}

Wire these from your Hono context (c.env) or Worker bindings.

Usage

Stream API

import {
  uploadFromUrl,
  getStreamVideo,
  getStreamEmbedUrl,
  deleteStreamVideo,
} from '@latimer-woods-tech/video';

// Copy a rendered MP4 from R2 into Stream
const video = await uploadFromUrl(
  'https://pub.r2.dev/renders/job_01.mp4',
  { appId: 'prime_self', topic: 'Q4 launch' },
  env,
);

// Embed it on a landing page
const embedUrl = getStreamEmbedUrl(video.uid);

R2 bucket binding

import { putR2Object, getR2Object } from '@latimer-woods-tech/video';

// Store a rendered MP4
const key = await putR2Object(env.VIDEOS_BUCKET, 'renders/job_01.mp4', arrayBuffer);

// Retrieve it for further processing
const buffer = await getR2Object(env.VIDEOS_BUCKET, 'renders/job_01.mp4');

Render job type

import type { RenderJob } from '@latimer-woods-tech/video';

The RenderJob type flows through the entire pipeline: from the video_calendar schedule, through the GitHub Actions render workflow, and back into @latimer-woods-tech/schedule for status tracking.

Constraints

• No Node.js built-ins (fs, path, crypto) — fully platform-neutral
• No Buffer — uses ArrayBuffer and Uint8Array
• All fetch calls include error handling
• Zero any in public APIs (TypeScript strict)

License

Factory Core — internal use only.