georgejeffers

alchemy

Alchemy IaC framework for TypeScript. Use when the user mentions Alchemy, wants to set up infrastructure, deploy Cloudflare Workers, configure databases, KV, R2, queues, use bindings and secrets, set up dev mode, use framework adapters (Vite, Astro, React Router, SvelteKit, Nuxt, TanStack Start), create custom resources, or work with any Alchemy provider.

georgejeffers 0 Updated 3mo ago

Resources

1
GitHub

Install

npx skillscat add georgejeffers/alchemy-skills/alchemy

Install via the SkillsCat registry.

SKILL.md

Alchemy

Alchemy is a TypeScript-native Infrastructure-as-Code framework. This skill provides comprehensive knowledge of Alchemy's APIs, patterns, and conventions.

Load references/alchemy-concepts.md for full details on any topic below.

Quick Start

New Project

alchemy create my-app --template vite

Add to Existing Project

alchemy init

Core File: alchemy.run.ts

import alchemy from "alchemy";
import { Worker, KVNamespace } from "alchemy/cloudflare";

const app = await alchemy("my-app");

const kv = await KVNamespace("cache", { title: "my-cache" });

const worker = await Worker("api", {
  entrypoint: "./src/worker.ts",
  bindings: {
    CACHE: kv,
    API_KEY: alchemy.secret(process.env.API_KEY),
    STAGE: app.stage,
  },
});

await app.finalize();

Commands

alchemy deploy              # deploy (default stage = $USER)
alchemy deploy --stage prod # deploy to prod
alchemy dev                 # local dev with hot reload
alchemy destroy             # tear down all resources

Task Routing

Based on what the user asks, determine which knowledge to load:

User Request What to Do
Set up Alchemy in a project Use alchemy create or alchemy init, create alchemy.run.ts
Add a Worker import { Worker } from "alchemy/cloudflare", configure with entrypoint + bindings
Add a database/KV/R2/queue Import from alchemy/cloudflare, create resource, bind to Worker
Use secrets alchemy.secret(process.env.X) for input, Secret.unwrap() in custom resources
Set up dev mode alchemy dev, configure framework adapter/plugin
Use a framework (Vite/Astro/etc.) Load references/alchemy-concepts.md §11 for framework adapters
Deploy to production alchemy deploy --stage prod, see CI guide patterns
Use Neon/Stripe/AWS/other provider Load references/alchemy-concepts.md §13 for provider list
Build a custom resource Load references/resource-patterns.md for implementation patterns
Build a new provider for Alchemy Run the full Provider Development Workflow below

Key Concepts

Resources

Resources are memoized async functions with create/update/delete lifecycle. Every resource takes an ID and props:

const db = await D1Database("my-db", { name: "my-db" });

Bindings

Connect resources to Workers with type-safe bindings:

const worker = await Worker("api", {
  entrypoint: "./src/worker.ts",
  bindings: { DB: db, KV: kv, SECRET: alchemy.secret("value") },
});

Access in worker code with type inference:

import type { worker } from "../alchemy.run";
export default {
  async fetch(req: Request, env: typeof worker.Env) {
    await env.DB.prepare("SELECT 1").run();
  },
};

Secrets

alchemy.secret(process.env.API_KEY)  // wrap a value
alchemy.secret.env.API_KEY           // shorthand

Stages

Isolated copies of your infrastructure. Default is $USER locally:

alchemy deploy              # deploys to $USER stage
alchemy deploy --stage prod # deploys to prod stage

Dev Mode

Local emulation with Miniflare, hot reload, framework integration:

alchemy dev

Framework Adapters

Each framework has a resource and a Vite plugin/adapter:

Framework Resource Plugin/Adapter
Vite Vite alchemy/cloudflare/vite
Astro Astro alchemy/cloudflare/astro
React Router ReactRouter alchemy/cloudflare/react-router
SvelteKit SvelteKit alchemy/cloudflare/sveltekit
Nuxt Nuxt alchemy/cloudflare/nuxt
TanStack Start TanStackStart alchemy/cloudflare/tanstack-start
BunSPA BunSPA (no plugin needed)
Next.js Nextjs alchemy/cloudflare/nextjs
Redwood (RWSDK) Redwood alchemy/cloudflare/rwsdk

Providers

Alchemy has 20+ providers. The main ones:

Provider Import Key Resources
Cloudflare alchemy/cloudflare Worker, D1Database, KVNamespace, R2Bucket, Queue, DurableObjectNamespace, Hyperdrive, Zone, DnsRecords
Neon alchemy/neon NeonProject, NeonBranch, NeonDatabase, NeonRole
Stripe alchemy/stripe WebhookEndpoint, Price, Product, Coupon, and 10+ more
AWS alchemy/aws Function, Table, Vpc, Subnet, SecurityGroup, Bucket, Role
AWS Control alchemy/aws/control AWS.{Service}.{Resource} — covers full AWS CloudFormation
GitHub alchemy/github GitHubSecret
Vercel alchemy/vercel VercelProject, VercelDnsRecord, VercelDeployment

Conventions

  • Default import: import alchemy from "alchemy"
  • Named imports from subpaths: import { Worker } from "alchemy/cloudflare"
  • await app.finalize() is always the last line of alchemy.run.ts
  • Physical names include app name + stage for uniqueness
  • Stage defaults to $USER locally; use --stage prod for production
  • Sensitive values always wrapped with alchemy.secret()
  • State stored in .alchemy/ directory (add to .gitignore or commit for shared state)

Reference Files

Load these on demand for detailed information:

  • references/alchemy-concepts.md — Comprehensive reference covering all Alchemy concepts: apps, stages, scopes, phases, resources, secrets, state, bindings, CLI, dev mode, profiles, framework adapters, serialization, and all 20+ providers with their resources.
  • references/resource-patterns.md — Implementation patterns for building custom resources (17 sections). Load when the user wants to create custom resources or contribute to Alchemy.
  • references/test-patterns.md — Test conventions and patterns (15 sections). Load when writing tests for custom resources.
  • references/doc-patterns.md — Documentation templates (10 sections). Load when writing docs for Alchemy providers.
  • references/checklist.md — Completeness verification checklist. Load when verifying a provider implementation.

Provider Development Workflow

Use this workflow when the user is building a new provider or custom resource for the Alchemy framework itself (not just using Alchemy in their app).

Phase 1: Research

Study the provider's API, identify CRUD-capable entities, authentication method, rate limits.

Phase 2: Design

Define resource list, Props/Output interfaces, API client approach, dependency order.

Phase 3: API Client

Create alchemy/src/{provider}/api.ts — minimal fetch wrapper, env var auth with Secret.unwrap() fallback.

Phase 4: Resources

Implement each resource. Load references/resource-patterns.md.

  • Resource("{provider}::{Resource}", ...) with create/update/delete lifecycle
  • Physical name: props.name ?? this.output?.name ?? this.scope.createPhysicalName(id)
  • Must use function declaration (not arrow), context via this
  • Export type guard (is{Resource}) for binding support

Phase 5: Tests

Write tests. Load references/test-patterns.md.

  • alchemy.test(import.meta, { prefix: BRANCH_PREFIX })
  • try/finally with destroy(scope) and API verification

Phase 6: Documentation

Write docs, guide, examples. Load references/doc-patterns.md.

  • Resource docs: alchemy-web/src/content/docs/providers/{provider}/{resource}.md
  • Guide: alchemy-web/src/content/docs/guides/{provider}.mdx
  • Example: examples/{provider}/alchemy.run.ts

Phase 7: Verify

Run checklist. Load references/checklist.md.

  • bun format, bun vitest alchemy/test/{provider}/, bun tsc -b