create-creek-app

Scaffold a new Creek project from a template — deploy to the edge in two commands.

What is Creek?

Creek is a deployment platform built on Cloudflare Workers. It replaces the manual setup of wrangler.toml, bindings, and resource provisioning with a single creek.toml config and zero-config CLI.

• creek.toml replaces wrangler.toml — Creek manages Workers, routes, and bindings for you
• creek deploy replaces wrangler deploy — one command provisions D1 databases, KV namespaces, R2 buckets, and deploys your code
• You own the Cloudflare account — Creek deploys to your CF account via OAuth, no vendor lock-in

If you already use Wrangler, Creek is a higher-level abstraction. If you're new to Cloudflare, Creek is the fastest way to get started.

Prerequisites

• Node.js 18+ (LTS recommended)
• Cloudflare account — free tier works. Not needed for scaffolding, only for creek deploy (you'll be prompted to sign in via OAuth on first deploy)

creek vs create-creek-app: create-creek-app scaffolds a new project. creek is the CLI that deploys, runs dev server, and manages your project. Both are installed on-demand via npx.

Quick Start

# 1. Scaffold a landing page
npx create-creek-app my-site --template landing --yes

# 2. Local development
cd my-site && npx creek dev
# → http://localhost:5173 with hot reload

# 3. Deploy to production
npx creek deploy
# → https://my-site-your-team.bycreek.com (live in ~10 seconds)

You get a globally distributed site on Cloudflare's edge network, with optional D1 database, KV storage, R2 files, AI inference, and Realtime WebSockets — all managed by Creek.

Usage

Interactive (human)

npx create-creek-app
# prompts: template → project name → scaffold → install → git init

Non-interactive (agent / CI)

npx create-creek-app my-blog --template blog --data '{"name":"Alice"}' --yes

From a JSON config file

npx create-creek-app my-site --template landing --data-file config.json --yes

// config.json
{
  "title": "Acme Corp",
  "tagline": "The best widgets",
  "theme": "light",
  "features": [
    { "title": "Fast", "description": "Edge-powered performance" },
    { "title": "Secure", "description": "Built-in DDoS protection" }
  ]
}

Template discovery

# List all templates
npx create-creek-app --list

[
  { "name": "blank", "description": "Minimal Creek project (no UI)", "capabilities": [] },
  { "name": "landing", "description": "Landing page with hero and CTA", "capabilities": [] },
  { "name": "blog", "description": "Blog with posts (D1 database)", "capabilities": ["d1"] },
  ...
]

# Print a template's JSON Schema
npx create-creek-app --template landing --schema

{
  "$schema": "https://json-schema.org/draft/2020-12/schema",
  "type": "object",
  "properties": {
    "title": { "type": "string", "default": "My Product" },
    "theme": { "type": "string", "enum": ["light", "dark"], "default": "dark" }
  }
}

# Validate data before scaffolding
npx create-creek-app --template landing --validate --data '{"theme":"dark"}'
# → { "valid": true, "errors": [] }

Third-party templates

npx create-creek-app --template github:user/my-template

Options

npx create-creek-app [dir] [options]

  dir                                     Project directory (default: prompted or "my-creek-app")

  -t, --template <name|github:user/repo>  Template to use
  --data <json>                           JSON data for template params
  --data-file <path>                      JSON file for template params
  --list                                  List available templates (JSON)
  --schema                                Print template JSON Schema
  --validate                              Validate data against schema
  --registry <url>                        Private template registry (enterprise)
  -y, --yes                               Skip prompts, use defaults
  --no-install                            Skip dependency installation
  --no-git                                Skip git init

Templates

Each template is a complete, deployable project — not a skeleton. Capabilities indicate which Creek-managed resources are pre-configured:

| Template | Description | Capabilities | |----------|-------------|-------------| | blank | Minimal Creek project (no UI) | — | | landing | Landing page with hero and CTA | — | | blog | Blog with posts | D1 database | | link-in-bio | Social links page | — | | api | REST API with Hono | D1 database | | todo | Realtime todo app | D1 + WebSocket | | dashboard | Data dashboard | D1 + WebSocket | | form | Form collector | D1 database | | chatbot | AI chatbot | D1 + Workers AI |

Capabilities explained:

• D1 — SQLite database at the edge, managed by Creek (no connection strings, no migrations)
• Realtime — WebSocket connections via Creek's Realtime service (presence, broadcast)
• AI — Inference via Cloudflare Workers AI (LLMs, embeddings, image generation)

Template Schema

Each template defines customizable parameters via JSON Schema in creek-template.json. This enables both human prompts and programmatic validation.

creek-template.json

{
  "name": "landing",
  "description": "Landing page with hero, features, and CTA",
  "capabilities": [],
  "schema": {
    "$schema": "https://json-schema.org/draft/2020-12/schema",
    "type": "object",
    "properties": {
      "title": { "type": "string", "default": "My Product" },
      "tagline": { "type": "string", "default": "Ship faster with Creek" },
      "theme": {
        "type": "string",
        "enum": ["light", "dark"],
        "default": "dark"
      }
    }
  }
}

Fields

| Field | Type | Required | Description | |-------|------|----------|-------------| | name | string | yes | Template identifier | | description | string | yes | Short description | | capabilities | string[] | yes | Creek resources used: d1, kv, r2, ai, realtime | | thumbnail | string | no | Path or URL to thumbnail image (400x300 recommended) | | screenshot | string | no | Path or URL to full screenshot image | | schema | object | no | JSON Schema defining customizable parameters |

creek-data.json

Default values for the schema parameters. Your app reads this file at runtime — no build-time string replacement. Users customize their project by editing this one file.

{
  "title": "My Product",
  "tagline": "Ship faster with Creek",
  "theme": "dark"
}

// src/App.tsx — reads config at runtime
import data from "../creek-data.json";

export function App() {
  return <h1>{data.title}</h1>;
}

Validation

Data is validated against the template's JSON Schema using ajv. Validation runs automatically during scaffold and can be triggered independently with --validate.

Supported constraints

• type — string, number, integer, boolean, array, object
• enum — fixed set of allowed values
• default — applied when the property is omitted
• required — within object items (array entries, nested objects)
• items — schema for array elements
• nested objects — full recursive validation

Validation examples

# Valid — passes (exit code 0)
npx create-creek-app --template landing --validate --data '{"theme":"dark"}'
# → { "valid": true, "errors": [] }

# Invalid enum — fails (exit code 1)
npx create-creek-app --template landing --validate --data '{"theme":"invalid"}'
# → { "valid": false, "errors": [{ "path": "/theme", "message": "must be equal to one of the allowed values" }] }

Exit Codes

| Code | Meaning | |------|---------| | 0 | Success | | 1 | Error — validation failure, invalid arguments, scaffold failure, or network error |

All structured output (--list, --schema, --validate) goes to stdout as JSON. Progress messages and errors go to stderr.

Creating Custom Templates

A Creek template is a directory with at minimum a package.json and creek.toml. Add creek-template.json for schema-driven customization.

Template structure

my-template/
├── creek-template.json       ← schema + metadata (removed on scaffold)
├── creek-data.json           ← default values (optional)
├── creek.toml                ← Creek project config
├── package.json
├── src/                      ← your code
├── worker/index.ts           ← edge worker (if needed)
└── .gitignore

Best practices

1. Use JSON Schema defaults — every property should have a default so the template works without any --data
2. Read config at runtime — import creek-data.json in your code instead of build-time placeholders
3. Keep schemas flat — prefer top-level string/number/boolean properties
4. Use enum for constrained choices — themes, layouts, color schemes. Agents discover valid options via --schema
5. Include a name property — create-creek-app auto-populates it from the project directory

Publishing

Host your template on GitHub and use it directly:

npx create-creek-app --template github:yourname/my-template

Or submit it to the Creek template gallery via pull request.

Agent Workflow

create-creek-app is designed for both humans and AI agents. Typical agent flow:

# 1. Discover templates
npx create-creek-app --list
# → JSON array to stdout

# 2. Read schema for chosen template
npx create-creek-app --template landing --schema
# → JSON Schema to stdout

# 3. Validate generated data
npx create-creek-app --template landing --validate --data '{"title":"Acme","theme":"dark"}'
# → { "valid": true, "errors": [] }

# 4. Scaffold
npx create-creek-app my-site --template landing --data '{"title":"Acme"}' --yes

# 5. Deploy
cd my-site && npx creek deploy --yes
# → { "ok": true, "url": "https://my-site-team.bycreek.com", ... }

All discovery and validation commands output JSON to stdout for programmatic consumption. Use --yes to skip all interactive prompts.

Contributing Templates

The template ecosystem is open to everyone. We welcome contributions from individual developers, agencies, and framework authors.

How to submit

1. Fork solcreek/templates
2. Create your template directory following the template structure
3. Ensure it scaffolds and deploys cleanly: npx create-creek-app test-dir --template ./your-template --yes && cd test-dir && creek deploy
4. Open a pull request

Acceptance criteria

• Works out of the box — --yes with no --data must produce a deployable project
• Has creek-template.json — with description, capabilities, and schema (if customizable)
• No secrets or API keys — templates must not require external service credentials to scaffold
• English — template names, descriptions, and code comments in English

Template tiers

| Tier | Badge | Maintained by | Review | |------|-------|--------------|--------| | Official | solcreek/templates | Creek team | Creek team review | | Community | github:author/repo | Author | Self-published, no review required |

Official templates are held to a higher standard: they must stay compatible with the latest Creek CLI and are tested in CI. Community templates can be used by anyone via --template github:author/repo — no approval process needed.

Ideas for templates

We'd especially love to see:

• Framework-specific starters (Astro + Creek, SvelteKit + Creek)
• Industry verticals (restaurant menu, event page, portfolio)
• Backend patterns (webhook receiver, cron job, queue processor)

See the templates repo for the full contribution guide.

License

Apache-2.0