create-creek-app 0.1.0 → 0.2.0

package/AGENTS.md

@@ -0,0 +1,136 @@
+# create-creek-app — Agent Reference
+
+You are interacting with `create-creek-app`, a CLI that scaffolds Creek projects from templates. Creek deploys to Cloudflare Workers.
+
+## Capabilities
+
+- Scaffold a new project from a built-in or third-party template
+- Discover available templates and their schemas (JSON output)
+- Validate template parameters before scaffolding
+- Customize templates via JSON data (inline or file)
+
+## Command Reference
+
+All commands output JSON to **stdout**. Progress/errors go to **stderr**.
+
+### Discovery (read-only, no side effects)
+
+```bash
+npx create-creek-app --list
+```
+Returns JSON array to stdout:
+```json
+[
+  { "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"] }
+]
+```
+
+```bash
+npx create-creek-app --template landing --schema
+```
+Returns the template's JSON Schema to stdout:
+```json
+{
+  "$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" }
+  }
+}
+```
+
+```bash
+npx create-creek-app --template landing --validate --data '{"theme":"dark"}'
+```
+Returns validation result to stdout:
+```json
+{ "valid": true, "errors": [] }
+```
+On failure:
+```json
+{ "valid": false, "errors": [{ "path": "/theme", "message": "must be equal to one of the allowed values" }] }
+```
+
+### Scaffold (creates files)
+
+```bash
+# Non-interactive scaffold
+npx create-creek-app <dir> --template <name> --yes
+
+# With custom parameters
+npx create-creek-app <dir> --template <name> --data '<json>' --yes
+
+# From a JSON config file
+npx create-creek-app <dir> --template <name> --data-file <path> --yes
+
+# Skip install and git init (faster, useful for testing)
+npx create-creek-app <dir> --template <name> --yes --no-install --no-git
+
+# Third-party template from GitHub
+npx create-creek-app <dir> --template github:user/repo --yes
+```
+
+### Deploy (after scaffolding)
+
+```bash
+cd <dir> && npx creek deploy --yes
+```
+
+## Recommended Workflow
+
+1. **`--list`** → pick a template based on description and capabilities
+2. **`--schema`** → read the JSON Schema to know valid parameters
+3. **`--validate`** → check your generated data (optional, scaffold validates too)
+4. **Scaffold** with `--template`, `--data`, `--yes`
+5. **Deploy** with `npx creek deploy --yes`
+
+## Exit Codes
+
+| Code | Meaning |
+|------|---------|
+| `0` | Success |
+| `1` | Error (validation failure, bad arguments, network error) |
+
+## Available Templates
+
+| Name | Capabilities | Use when |
+|------|-------------|----------|
+| `blank` | — | Agent generates all code from scratch |
+| `landing` | — | Marketing/product landing page |
+| `blog` | D1 | Content site with database |
+| `link-in-bio` | — | Social links page |
+| `api` | D1 | REST API backend |
+| `todo` | D1, Realtime | Full-stack app with WebSocket |
+| `dashboard` | D1, Realtime | Data visualization with live updates |
+| `form` | D1 | Form submissions with database storage |
+| `chatbot` | D1, AI | AI-powered chat with conversation history |
+
+## Template Parameters
+
+Parameters are template-specific. Always run `--schema` to discover them. Common patterns:
+
+- `name` — project name (auto-populated from directory name)
+- `title`, `tagline`, `description` — display text
+- `theme` — typically `"light"` or `"dark"` (enum-constrained)
+- `accentColor` — hex color string
+- `features` — array of `{ title, description }` objects
+
+## What Creek Provides After Deploy
+
+The scaffolded project deploys to Cloudflare's edge network via `creek deploy`. Depending on template capabilities:
+
+- **All templates**: Global CDN, HTTPS, preview URLs
+- **D1**: SQLite database — use `import { db } from "creek"` (no connection strings)
+- **Realtime**: WebSocket — use `import { usePresence } from "creek/react"`
+- **AI**: Workers AI — use `import { ai } from "creek"`
+
+## Important Notes
+
+- Always use `--yes` to skip interactive prompts
+- The `--data` flag accepts inline JSON; use `--data-file` for complex data
+- Templates use runtime config (`creek-data.json`) — users can customize after scaffold by editing this file
+- `creek-template.json` is metadata only — it's removed from the scaffolded project
+- No API keys or credentials needed for scaffolding; `creek deploy` handles auth via OAuth

package/README.md

@@ -1,11 +1,44 @@
 # create-creek-app
 
-Scaffold a new [Creek](https://creek.dev) project from a template.
+[![npm version](https://img.shields.io/npm/v/create-creek-app)](https://www.npmjs.com/package/create-creek-app)
+[![license](https://img.shields.io/npm/l/create-creek-app)](https://github.com/solcreek/creek/blob/main/LICENSE)
+
+Scaffold a new [Creek](https://creek.dev) project from a template — deploy to the edge in two commands.
+
+## What is Creek?
+
+Creek is a deployment platform built on [Cloudflare Workers](https://developers.cloudflare.com/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
 
 ```bash
-npx create-creek-app
+# 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)
@@ -21,17 +54,61 @@ npx create-creek-app
 npx create-creek-app my-blog --template blog --data '{"name":"Alice"}' --yes
 ```
 
+### From a JSON config file
+
+```bash
+npx create-creek-app my-site --template landing --data-file config.json --yes
+```
+
+```json
+// 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
 
 ```bash
-# List all templates (JSON)
+# List all templates
 npx create-creek-app --list
+```
 
+```json
+[
+  { "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"] },
+  ...
+]
+```
+
+```bash
 # Print a template's JSON Schema
 npx create-creek-app --template landing --schema
+```
 
+```json
+{
+  "$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" }
+  }
+}
+```
+
+```bash
 # Validate data before scaffolding
 npx create-creek-app --template landing --validate --data '{"theme":"dark"}'
+# → { "valid": true, "errors": [] }
 ```
 
 ### Third-party templates
@@ -45,7 +122,7 @@ npx create-creek-app --template github:user/my-template
 ```
 npx create-creek-app [dir] [options]
 
-  dir                                   Project directory (default: prompted or "my-creek-app")
+  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
@@ -61,17 +138,24 @@ npx create-creek-app [dir] [options]
 
 ## 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 |
+| `blog` | Blog with posts | D1 database |
 | `link-in-bio` | Social links page | — |
-| `api` | REST API with Hono | D1 |
-| `todo` | Realtime todo app | D1, Realtime |
-| `dashboard` | Data dashboard | D1, Realtime |
-| `form` | Form collector | D1 |
-| `chatbot` | AI chatbot | D1, AI |
+| `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
 
@@ -84,8 +168,6 @@ Each template defines customizable parameters via [JSON Schema](https://json-sch
   "name": "landing",
   "description": "Landing page with hero, features, and CTA",
   "capabilities": [],
-  "thumbnail": "thumbnail.png",
-  "screenshot": "screenshot.png",
   "schema": {
     "$schema": "https://json-schema.org/draft/2020-12/schema",
     "type": "object",
@@ -108,14 +190,14 @@ Each template defines customizable parameters via [JSON Schema](https://json-sch
 |-------|------|----------|-------------|
 | `name` | string | yes | Template identifier |
 | `description` | string | yes | Short description |
-| `capabilities` | string[] | yes | Required Creek resources: `d1`, `kv`, `r2`, `ai`, `realtime` |
-| `thumbnail` | string | no | Relative path to thumbnail image (400x300 recommended) |
-| `screenshot` | string | no | Relative path to full screenshot image |
+| `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. The app reads this file at runtime — no build-time string replacement.
+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.
 
 ```json
 {
@@ -125,6 +207,15 @@ Default values for the schema parameters. The app reads this file at runtime —
 }
 ```
 
+```tsx
+// 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](https://ajv.js.org/). Validation runs automatically during scaffold and can be triggered independently with `--validate`.
@@ -141,19 +232,24 @@ Data is validated against the template's JSON Schema using [ajv](https://ajv.js.
 ### Validation examples
 
 ```bash
-# Valid — passes
+# Valid — passes (exit code 0)
 npx create-creek-app --template landing --validate --data '{"theme":"dark"}'
 # → { "valid": true, "errors": [] }
 
-# Invalid enum — fails
+# 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" }] }
-
-# Wrong type — fails
-npx create-creek-app --template landing --validate --data '{"title":123}'
-# → { "valid": false, "errors": [{ "path": "/title", "message": "must be string" }] }
 ```
 
+## 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.
@@ -162,25 +258,22 @@ A Creek template is a directory with at minimum a `package.json` and `creek.toml
 
 ```
 my-template/
-├── creek-template.json       ← schema + metadata (optional)
+├── 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                ← renamed to .gitignore on scaffold
+└── .gitignore
 ```
 
 ### Best practices
 
-1. **Use JSON Schema defaults** — every property in the schema 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 using build-time placeholders. This lets users customize after scaffold by editing one file
-3. **Keep schemas flat** — prefer top-level string/number/boolean properties. Use arrays only for naturally repeated structures (e.g., feature lists)
-4. **Use `enum` for constrained choices** — themes, layouts, color schemes. Agents can discover valid options via `--schema`
-5. **Name the `name` property** — include a `name` property in your schema. `create-creek-app` auto-populates it from the project directory
-6. **Add a thumbnail** — 400x300 PNG for template gallery display. Optional, but recommended for visual templates
-7. **Add a screenshot** — full-page screenshot showing the template in action. Optional
-8. **Use `_gitignore`** — npm strips `.gitignore` from published packages. Use `_gitignore` and it will be renamed automatically
+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
 
@@ -199,21 +292,61 @@ Or submit it to the [Creek template gallery](https://github.com/solcreek/templat
 ```bash
 # 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
+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.
+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](https://github.com/solcreek/templates)
+2. Create your template directory following the [template structure](#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](https://github.com/solcreek/templates) for the full contribution guide.
 
 ## License
 

package/dist/scaffold.js

@@ -1,5 +1,5 @@
 import { resolve, basename } from "node:path";
-import { existsSync, renameSync, unlinkSync } from "node:fs";
+import { existsSync, unlinkSync } from "node:fs";
 import { execSync } from "node:child_process";
 import consola from "consola";
 import { fetchTemplate } from "./fetch.js";
@@ -26,13 +26,7 @@ export async function scaffold(opts) {
     }
     // 3. Apply data
     applyData(dir, userData, defaultData);
-    // 4. Rename _gitignore → .gitignore
-    const gitignoreSrc = resolve(dir, "_gitignore");
-    const gitignoreDest = resolve(dir, ".gitignore");
-    if (existsSync(gitignoreSrc)) {
-        renameSync(gitignoreSrc, gitignoreDest);
-    }
-    // 5. Remove creek-template.json from output (metadata, not project file)
+    // 4. Remove creek-template.json from output (metadata, not project file)
     const templateConfigPath = resolve(dir, "creek-template.json");
     if (existsSync(templateConfigPath)) {
         unlinkSync(templateConfigPath);

package/dist/templates.d.ts

@@ -1,39 +1,19 @@
-/** Hardcoded template list — used for --list and interactive prompt. */
-export declare const TEMPLATES: readonly [{
-    readonly name: "blank";
-    readonly description: "Minimal Creek project (no UI)";
-    readonly capabilities: string[];
-}, {
-    readonly name: "landing";
-    readonly description: "Landing page with hero and CTA";
-    readonly capabilities: string[];
-}, {
-    readonly name: "blog";
-    readonly description: "Blog with posts (D1 database)";
-    readonly capabilities: readonly ["d1"];
-}, {
-    readonly name: "link-in-bio";
-    readonly description: "Social links page";
-    readonly capabilities: string[];
-}, {
-    readonly name: "api";
-    readonly description: "REST API with Hono (D1)";
-    readonly capabilities: readonly ["d1"];
-}, {
-    readonly name: "todo";
-    readonly description: "Realtime todo app (D1 + WebSocket)";
-    readonly capabilities: readonly ["d1", "realtime"];
-}, {
-    readonly name: "dashboard";
-    readonly description: "Data dashboard (D1 + Realtime)";
-    readonly capabilities: readonly ["d1", "realtime"];
-}, {
-    readonly name: "form";
-    readonly description: "Form collector (D1)";
-    readonly capabilities: readonly ["d1"];
-}, {
-    readonly name: "chatbot";
-    readonly description: "AI chatbot (D1 + Workers AI)";
-    readonly capabilities: readonly ["d1", "ai"];
-}];
+/**
+ * Template catalog — used for --list, interactive prompt, and gallery.
+ *
+ * Four types by output:
+ *   site      — Visual pages (deploy → shareable URL)
+ *   app       — Interactive UI + backend logic
+ *   workflow  — Background process (no/minimal UI), trigger + steps
+ *   connector — Data bridge pattern (ingestion → storage)
+ */
+export interface Template {
+    name: string;
+    description: string;
+    type: "site" | "app" | "workflow" | "connector" | "developer";
+    capabilities: string[];
+    trigger?: string;
+}
+export declare const TEMPLATES: Template[];
 export type TemplateName = (typeof TEMPLATES)[number]["name"];
+export type TemplateType = Template["type"];

package/dist/templates.js

@@ -1,12 +1,38 @@
-/** Hardcoded template list — used for --list and interactive prompt. */
+/**
+ * Template catalog — used for --list, interactive prompt, and gallery.
+ *
+ * Four types by output:
+ *   site      — Visual pages (deploy → shareable URL)
+ *   app       — Interactive UI + backend logic
+ *   workflow  — Background process (no/minimal UI), trigger + steps
+ *   connector — Data bridge pattern (ingestion → storage)
+ */
 export const TEMPLATES = [
-    { 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"] },
-    { name: "link-in-bio", description: "Social links page", capabilities: [] },
-    { name: "api", description: "REST API with Hono (D1)", capabilities: ["d1"] },
-    { name: "todo", description: "Realtime todo app (D1 + WebSocket)", capabilities: ["d1", "realtime"] },
-    { name: "dashboard", description: "Data dashboard (D1 + Realtime)", capabilities: ["d1", "realtime"] },
-    { name: "form", description: "Form collector (D1)", capabilities: ["d1"] },
-    { name: "chatbot", description: "AI chatbot (D1 + Workers AI)", capabilities: ["d1", "ai"] },
+    // --- Sites (Level 4 — deploy → shareable) ---
+    { name: "landing", description: "Landing page with hero and CTA", type: "site", capabilities: [] },
+    { name: "blog", description: "Blog with image uploads", type: "site", capabilities: ["database", "storage"] },
+    { name: "link-in-bio", description: "Social links page", type: "site", capabilities: [] },
+    { name: "waitlist", description: "Waitlist with AI user segmentation", type: "site", capabilities: ["database", "ai"] },
+    // --- Apps (Level 3 — working data pipeline + UI) ---
+    { name: "form", description: "Form collector with admin dashboard", type: "app", capabilities: ["database"] },
+    { name: "dashboard", description: "Data dashboard with scheduled refresh", type: "app", capabilities: ["database", "realtime", "cron"] },
+    { name: "chatbot", description: "AI chatbot with conversation history", type: "app", capabilities: ["database", "ai"] },
+    { name: "survey", description: "Survey with AI-powered analysis", type: "app", capabilities: ["database", "ai"] },
+    { name: "knowledge-base", description: "Knowledge base with AI search", type: "app", capabilities: ["database", "ai", "storage"] },
+    { name: "status-page", description: "Service status page with uptime monitoring", type: "app", capabilities: ["database", "realtime", "cron", "cache"] },
+    { name: "file-share", description: "File upload/download with expiring links", type: "app", capabilities: ["database", "storage"] },
+    { name: "todo", description: "Realtime todo app (learning example)", type: "app", capabilities: ["database", "realtime"] },
+    // --- Workflows (Level 2-3 — trigger + steps, no/minimal UI) ---
+    { name: "approval-flow", description: "Approval workflow — leave, expense, procurement", type: "workflow", capabilities: ["database", "realtime", "queue"], trigger: "webhook" },
+    { name: "invoice-processor", description: "Email invoice → AI extract → DB → notify", type: "workflow", capabilities: ["database", "ai", "storage", "email"], trigger: "email" },
+    { name: "scheduled-report", description: "Scheduled query → AI summary → PDF report", type: "workflow", capabilities: ["database", "ai", "storage", "cron"], trigger: "cron" },
+    { name: "data-sync", description: "Scheduled API pull → diff → DB → notify", type: "workflow", capabilities: ["database", "cron"], trigger: "cron" },
+    { name: "ai-classifier", description: "Input → AI classify → route to queue", type: "workflow", capabilities: ["database", "ai", "queue"], trigger: "webhook" },
+    // --- Connectors (Level 2 — pure platform wiring) ---
+    { name: "api", description: "REST API with Hono", type: "connector", capabilities: ["database"], trigger: "http" },
+    { name: "webhook-receiver", description: "Receive and process external webhooks", type: "connector", capabilities: ["database", "queue"], trigger: "webhook" },
+    { name: "email-to-db", description: "Email → parse → attachments to R2 → DB", type: "connector", capabilities: ["database", "storage", "email"], trigger: "email" },
+    { name: "ftp-sync", description: "Scheduled FTP/SFTP pull → parse → DB", type: "connector", capabilities: ["database", "storage", "cron", "ftp"], trigger: "cron" },
+    // --- Developer ---
+    { name: "blank", description: "Minimal Creek project (empty canvas)", type: "developer", capabilities: [] },
 ];

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "create-creek-app",
-  "version": "0.1.0",
+  "version": "0.2.0",
   "description": "Create a new Creek project from a template",
   "type": "module",
   "bin": {
@@ -8,7 +8,8 @@
   },
   "main": "./dist/index.js",
   "files": [
-    "dist"
+    "dist",
+    "AGENTS.md"
   ],
   "scripts": {
     "build": "tsc",