BigCommerce x Cloudflare Hackathon Kit
GitHub

Commerce agents, edge AI, and durable workflows

Eight BigCommerce demos for shipping Cloudflare-native commerce ideas fast.

A mentor-friendly hub for hackathon teams building storefront generators, voice shopping, returns automation, photo studios, analytics, catalog enrichment, and A/B testing on Cloudflare Workers.

Browse demosOpen kickoff deckView plan

Demo grid

All eight demo tracks in one operator view

Every demo is a standalone Cloudflare Worker TypeScript app with live-capable service seams, Workers AI runtime endpoints where AI belongs, and fixture fallbacks for reliability.

01 Tier 1

Product Photo Studio

Generate campaign-ready product photos through a Workers AI runtime endpoint, store variants in R2, serve optimized crops, and publish the selected asset back to a catalog path.

Workers AI endpoint ready
WorkersWorkers AIR2ImagesKV

BigCommerce path: Storefront product reads feed the image prompt; Catalog REST publishes when Admin credentials are present, with an activity-log fallback.

  1. Select product
  2. POST /api/generate-image
  3. Store variants
  4. Publish selected image
https://bc-photo-studio.dwarven.workers.dev https://bc-photo-studio.cloudflare.app
GitHubworkers.devcloudflare.app
02 Tier 1

Storefront Builder

Turn a merchant campaign prompt into generated storefront routes, product-aware components, a Sandbox preview, and deploy activity using Cloudflare-native Workers AI generation.

Workers AI builder endpoint ready
WorkersSandbox SDKWorkers AIStorefront GraphQL

BigCommerce path: Storefront GraphQL context feeds generation when credentials are present; fixture catalog context remains available as a fallback.

  1. Prompt campaign
  2. POST /api/generate
  3. Run Sandbox preview
  4. Deploy or export
https://bc-vibe-storefront.dwarven.workers.dev https://bc-vibe-storefront.cloudflare.app
GitHubworkers.devcloudflare.app
03 Tier 1

Voice Shopping Agent

A phone-style multilingual shopping agent that shows speech capture, STT, product tools, cart updates, and TTS response timing.

Polished mock voice flow
Agents SDKDurable ObjectsRealtimeKitWorkers AI

BigCommerce path: Storefront product, recommendation, and cart tools backed by fixtures until real Storefront credentials are configured.

  1. Speak intent
  2. Transcribe
  3. Recommend products
  4. Add to cart
https://bc-voice-shopping.dwarven.workers.dev https://bc-voice-shopping.cloudflare.app
GitHubworkers.devcloudflare.app
04 Tier 1

Returns Agent

Classify damaged-product photos through a Workers AI vision endpoint, recommend refund or replacement actions, gate high-value refunds, and schedule follow-up work.

Workers AI vision endpoint ready
Agents SDKDurable ObjectsWorkers AI VisionR2Scheduling

BigCommerce path: Orders and refunds call Admin REST through bc-tools when credentials are present, with fallback review records for dry runs.

  1. Upload photo
  2. POST /api/returns/classify
  3. Decide action
  4. Approve and follow up
https://bc-returns-agent.dwarven.workers.dev https://bc-returns-agent.cloudflare.app
GitHubworkers.devcloudflare.app
05 Tier 2

Shopping Concierge

A lower-risk text shopping concierge with recommendation cards, preference memory, cart previews, and visible tool-call telemetry.

Functional text agent mock
WorkersAgents SDKDurable ObjectsKV or D1

BigCommerce path: Mock product and preference tools mirror the future bc-tools Storefront and Admin descriptors.

  1. Ask for gear
  2. Infer preferences
  3. Search products
  4. Preview cart
https://bc-shopping-concierge.dwarven.workers.dev https://bc-shopping-concierge.cloudflare.app
GitHubworkers.devcloudflare.app
06 Tier 2

Data Analyst

Ask questions about store performance, preview generated Python, and see reproducible KPI, chart, and table outputs.

Interactive analyst mock
WorkersSandbox SDKR2Workers AI

BigCommerce path: Fixture datasets today; bc-tools clients replace CSV mocks with real storefront/admin exports when credentials exist.

  1. Choose dataset
  2. Ask question
  3. Generate Python
  4. Persist artifacts
https://bc-data-analyst.dwarven.workers.dev https://bc-data-analyst.cloudflare.app
GitHubworkers.devcloudflare.app
07 Tier 2

Catalog Enricher

Bulk-generate SEO titles, meta descriptions, image alt text, and tags through a Workers AI enrichment endpoint with durable progress and merchant review.

Workers AI enrichment endpoint ready
WorkersAgents SDKWorkflowsWorkers AI

BigCommerce path: Catalog REST updates publish approved rows when Admin credentials are configured; fixture updates remain a fallback.

  1. Scan SKUs
  2. POST /api/catalog/enrich
  3. Validate QA
  4. Publish approved rows
https://bc-catalog-enricher.dwarven.workers.dev https://bc-catalog-enricher.cloudflare.app
GitHubworkers.devcloudflare.app
08 Tier 2

A/B Test Generator

Generate landing page variants, simulate screenshot capture, score a decision matrix, and recommend the strongest campaign concept.

Review-board scaffold
WorkersSandbox SDKBrowser RunWorkers AIR2

BigCommerce path: Mock product assumptions now; live Storefront and Admin clients feed variant generation, scoring, and publish hooks later.

  1. Enter brief
  2. Generate variants
  3. Capture screenshots
  4. Pick winner
https://bc-ab-test-generator.dwarven.workers.dev https://bc-ab-test-generator.cloudflare.app
GitHubworkers.devcloudflare.app

BigCommerce data strategy

Live-commerce ready, with fallback fixtures when credentials are unavailable

The kit keeps live BigCommerce paths explicit without blocking hackathon teams on credentials. Product, order, customer, cart, and catalog fixtures cover fallback demos, while bc-tools centralizes Storefront GraphQL and Admin REST access.

Read path

Use Storefront GraphQL for product browsing, recommendations, carts, analytics context, and storefront generation when BIGCOMMERCE_STOREFRONT_TOKEN is available. Fall back to fixtures for local demos.

Write path

Route catalog, refund, image, and SEO updates through Admin REST when USE_BIGCOMMERCE_MOCKS=false and credentials are configured with explicit mentor review. Otherwise, write to reviewable fallback logs.

Shared layer

bc-tools owns clients, fixtures, Zod-shaped tools, and swap points so demos do not each invent a BigCommerce integration.

Architecture map

How the hub, demos, Cloudflare services, and BigCommerce seams fit together

Use this map to route teams quickly from an idea to the right service combination and demo repo.

01

Hub Worker

One-page guide, demo routing, mentor notes, and deploy targets.

02

Demo Workers

Eight standalone TypeScript Workers, each live-capable and deployable.

03

Agents + AI

Agents SDK, Workers AI, Realtime, vision, TTS, and tool calling.

04

Execution + Artifacts

Sandbox SDK, Browser Run, R2, Images, Workflows, KV, and D1.

05

bc-tools

Shared fixtures, Storefront GraphQL, Admin REST, and agent-ready tool descriptors.

06

BigCommerce

Storefront/Admin credentials unlock live commerce paths; fixtures cover fallback scenarios.

Mentor cheat sheet

Keep teams moving, even without live credentials

The fastest path is to pick a demo, run it locally, identify the live endpoint, then add only the bindings needed for the team’s core differentiator.

Resources

Repos, docs, design references, and platform primers

These links are the canonical places to send builders when they need setup commands, API details, or the design language behind the kit.

Next steps

Deploy, capture proof, then prepare for public release.

  • Deploy each Worker to workers.dev under the pgx account.
  • Provision matching cloudflare.app URLs for the hub, kickoff deck, and eight demos.
  • Link bc-tools into demos once package publishing or local workspace wiring is finalized.
  • Add screenshots, short walkthrough videos, and public-release health files before flipping repos public.
Hub repoKickoff deck repo