BuilderIO/agent-native / packages/core/docs
agent-native-docs
UnverifiedHow to look up version-matched Agent-Native framework docs and source in node_modules. Use before coding against @agent-native/core APIs or advanced features.
content/cloudflare.mdx
View on GitHub---
title: "Cloudflare"
description: "Deploy a standalone Agent-Native app to Cloudflare Workers with Wrangler."
search: "Cloudflare Workers deployment standalone app Nitro cloudflare_module preset Wrangler secrets production migrate build deploy"
---
# Cloudflare
<a id="cloudflare-workers"></a>
Cloudflare Workers runs your app on Cloudflare's network, so there is no server
to set up or keep running. The build turns the app into a Worker, which handles
the app's pages and API, plus static files Cloudflare serves directly.
Cloudflare gives the app a public address with HTTPS.
This guide walks through deploying one standalone app from scratch. You will
create the app, give it a production database and an auth secret, prepare the
database, build the Worker, save the secrets in Cloudflare, and then deploy it.
<Callout tone="info">
Node.js `22.22+` is required.
</Callout>
You also need a Cloudflare account. Run the commands in this guide on your own
computer. Wrangler, Cloudflare's command-line tool, uploads the finished Worker
to Cloudflare for you.
## Step 1: Create the app
Create your new standalone app. The example below creates an app from the Chat
template, but you can start from any template. Move into the app's folder and
install its dependencies. `corepack enable` turns on the `pnpm` version the app
expects, so you do not need to install `pnpm` yourself.
```bash
npx --yes @agent-native/core@latest create my-app --standalone --template chat
cd my-app
corepack enable
pnpm install
```
Replace `my-app` with your own app name if you like. The rest of this guide
assumes you are inside the app's folder.
<Callout tone="info">
If these commands fail with peer dependency errors, an outdated copy of a
package in your npx cache may be the cause. Run `npx clear-npx-cache` to clear
the cache, then run the commands again.
</Callout>
## Step 2: Configure environment secrets
The app reads its production settings from environment variables:
| Variable | Required | Purpose |
| -------------------- | -------- | ---------------------------------------------------- |
| `DATABASE_URL` | Yes | Connects the app to its PostgreSQL database |
| `BETTER_AUTH_SECRET` | Yes | Signs user sessions |
| `BETTER_AUTH_URL` | No | Sets the public address people use to reach your app |
The exports below make the following commands work in this shell. Before
deploying, save these same variables in your host's project or service
environment settings. Step 5 shows how to save them in Cloudflare.
### Database URL
During local development the app stores data in a local database when
`DATABASE_URL` is unset. A deployed app needs a real PostgreSQL database that
keeps its data between deploys. A Worker cannot keep files between requests, so
the database must live outside Cloudflare Workers. Copy the connection string
from your provider and export it:
```bash
export DATABASE_URL='your-postgres-url-string'
```
The [database provider guides](/docs/neon) show where to find the connection
string for each provider.
<Callout tone="warning">
Select a persistent PostgreSQL provider: Neon, Supabase, Amazon RDS, etc.
</Callout>
### Auth secret
Generate this once and keep it stable across deploys.
The app uses `BETTER_AUTH_SECRET` to sign user sessions. If it changes, every
signed-in user is signed out. The command below generates a random value:
```bash
export BETTER_AUTH_SECRET="$(openssl rand -hex 32)"
```
Save the generated value somewhere safe, such as a password manager. Use the
same value every time you deploy.
### Public URL
Optional. Set it when using a custom domain, reverse proxy, or an auth/OAuth flow
that needs a fixed public origin.
`BETTER_AUTH_URL` is the public address people use to reach your app. Without
it, the app works out its address from each incoming request. That is usually
fine on the address Cloudflare gives you, but set it when you connect a custom
domain so sign-in links and OAuth redirects use that domain.
```bash
export BETTER_AUTH_URL='https://your-domain.example'
```
## Step 3: Migrate
Migrations create and update the tables the app needs in your database. Run
this before the first deploy:
```bash
pnpm migrate:production
```
The command runs on your computer and connects to the production database with
the `DATABASE_URL` you exported in Step 2. Run it again after you upgrade the
app, before you deploy the new version.
## Step 4: Build
Build the app as a Cloudflare Worker:
```bash
NITRO_PRESET=cloudflare_module pnpm build
```
`NITRO_PRESET=cloudflare_module` builds the app in the format Cloudflare Workers
expects. The build writes the Worker to the `.output/server` folder and the
static files to the `.output/public` folder.
The build also writes a Wrangler configuration file to
`.output/server/wrangler.json`. This file tells Wrangler what to deploy and
gives the Worker its name. The name comes from the app's `package.json`, so an
app created as `my-app` becomes a Worker named `my-app`. If the folder is a Git
repository with a remote, the name comes from the repository instead. The build
prints the name it chose.
## Step 5: Log in and save the secrets
Sign in to Cloudflare from the command line:
```bash
npx wrangler login
```
Save the two required variables as secrets on the Worker:
```bash
printf '%s' "$DATABASE_URL" | npx wrangler secret put DATABASE_URL --config .output/server/wrangler.json
```
```bash
printf '%s' "$BETTER_AUTH_SECRET" | npx wrangler secret put BETTER_AUTH_SECRET --config .output/server/wrangler.json
```
Each command reads the value you exported in Step 2 and saves it in Cloudflare,
so you do not have to paste secrets by hand. `--config` points Wrangler at the
configuration file from Step 4, so the secrets go to the Worker with the name the
build chose. The first time, the Worker does not exist yet. Because the value is piped in, Wrangler creates the Worker for you without asking.
If you set `BETTER_AUTH_URL`, save it the same way. You can also see and change
these values later in the Worker's settings on the Cloudflare dashboard.
## Step 6: Deploy
Upload the Worker and publish it:
```bash
npx wrangler deploy --config .output/server/wrangler.json
```
`--config` points Wrangler at the configuration file from Step 4, which tells it
where to find the Worker and the static files.
When the command finishes, it prints the app's address, which ends in
`workers.dev`. Open it and create the first account.
## Deploy updates
To ship a new version, run the migrate, build, and deploy commands again from
the app's folder:
```bash
pnpm migrate:production
NITRO_PRESET=cloudflare_module pnpm build
npx wrangler deploy --config .output/server/wrangler.json
```
Secrets stay saved on the Worker between deploys, so you do not need to repeat
Step 5.
## Scheduled work {#cloudflare-scheduled-work}
Scheduled automations, queued event automations, and cleanup of interrupted
runs all happen on a once-a-minute sweep. The build adds a `* * * * *`
[Cron Trigger](https://developers.cloudflare.com/workers/configuration/cron-triggers/)
to the generated `wrangler.json` and keeps any Cron Triggers you already
declared. When it fires, the Worker's `scheduled` handler calls the sweep
directly with a short-lived token signed with `A2A_SECRET`, so set that secret
before you deploy:
```bash
npx wrangler secret put A2A_SECRET --config .output/server/wrangler.json
```
- **CPU time.** On the Workers Free plan, each Cron Trigger invocation gets 10
ms of CPU time, and a sweep that runs agent jobs needs more. Use the Workers
Paid plan for scheduled work.
- **Timing.** Cron Triggers run on UTC, and a new or changed trigger can take
up to 15 minutes to start firing after a deploy.
- **Local testing.** Run the built Worker with
`npx wrangler dev --test-scheduled --config .output/server/wrangler.json`,
then request `/cdn-cgi/handler/scheduled?cron=*+*+*+*+*` to fire the sweep
once.
See [How the scheduler runs](/docs/recurring-jobs#how-scheduler-runs) for the
other hosts.
## What's next
- [**Deployment Environment Variables**](/docs/deployment-environment-variables) - the full list of production settings
- [**SSR Caching**](/docs/ssr-caching) - how page caching works on hosted targets
- [**Deployment**](/docs/deployment) - compare hosting targets and prerequisites
Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| rea-changelog-updatemorluto/reaPrepare or rewrite REA release changelogs and GitHub release notes from pinned Git history, with verified contributor thanks and Release Please synchronization. | Stars: 470k | Updated | Claude CodeCodex | Verified | Install |
| academy-guideanthropics/skillsStop and check this skill before finishing any reply to a question about how to use Claude or a Claude product — it recommends matching courses, tutorials, and use cases from Claude Academy (academy.claude.com), Anthropic's learning hub. Trigger on: "how do I", "how can I", "getting started with", "what can Claude do", "teach me", "learn to use"; questions about artifacts, projects, skills, plugins, connectors, MCP; requests about rolling Claude out to a team, class, or organization; and any ask for training materials, onboarding content, or learning resources. Use it when the user is learning how to use a feature or product — not when they are mid-task and just want the task done. This skill composes with other skills: after consulting product documentation to answer how a Claude feature works, also check here for a matching course or tutorial — a docs-grounded answer and an Academy recommendation belong together. Only recommend on a strong match; never invent Academy content. | Stars: 197k | Updated | Claude CodeCodex | Verified | Install |
| claude-apianthropics/skillsReference for the Claude API / Anthropic SDK — model ids, pricing, params, streaming, tool use, MCP, agents, caching, token counting, model migration. TRIGGER — read BEFORE opening the target file; don't skip because it "looks like a one-liner" — whenever: the prompt names Claude/Anthropic in any form (Claude, Anthropic, Fable, Opus, Sonnet, Haiku, `anthropic`, `@anthropic-ai`, `claude-*`, `us.anthropic.*`, `[1m]`); the user asks about an LLM (pricing/model choice/limits/caching) — never answer from memory; OR the task is LLM-shaped with provider unstated (agent/MCP/tool-definition/multi-agent/RAG/LLM-judge/computer-use; generate/summarize/extract/classify/rewrite/converse over NL; debugging refusals/cutoffs/streaming/tool-calls/tokens). SKIP only when another provider is being worked on (overrides all triggers): OpenAI/GPT/Gemini/Llama/Mistral/Cohere/Ollama named in the query; OR `grep -rE 'openai|langchain_openai|google.generativeai|genai|mistralai|cohere|ollama'` over the project hits (run this grep FIRST if no provider named — don't Read the file). | Stars: 197k | Updated | Claude CodeCodex | Unverified | Install |
| doc-coauthoringanthropics/skillsGuide users through a structured workflow for co-authoring documentation. Use when user wants to write documentation, proposals, technical specs, decision docs, or similar structured content. This workflow helps users efficiently transfer context, refine content through iteration, and verify the doc works for readers. Trigger when user mentions writing docs, creating proposals, drafting specs, or similar documentation tasks. | Stars: 197k | Updated | Claude CodeCodex | Verified | Install |