How to self-host a Formspree alternative on Cloudflare

Inspect the pinned deployment before running it

Open the FormZero project page and follow its pinned repository links. The reviewed commit is a4ddde255457d1e6ddf55b5f24b51809d18501f3. Its Wrangler file points to workers/app.ts, enables Node compatibility and declares D1 as DB. Replace the repository's database ID with a database owned by your account. Keep the pilot separate from any existing production data.

The package's deploy script builds with React Router, applies migrations with --remote and then runs Wrangler deploy. It is a remote mutation, not merely a local preview. Inspect the SQL and intended binding first; create a backup before applying migrations to a populated database. This guide did not run the upstream script.

Follow the actual submission path

The Worker delegates requests to the React Router application. The submission route looks up a form in D1, parses JSON or form data and inserts a submission record. For JSON clients it returns a success response; HTML clients use a redirect path. The reviewed configuration does not add KV, R2 or a Queue merely because those services are available on Cloudflare.

Layer	Evidence	Pilot check
Browser form	HTML/JSON parsing in the route	Match the content type, expected response and thank-you destination
Worker	React Router request handler	Check build output and correct deployment binding
D1	DB and SQL migrations	Confirm tables, ownership and access with two synthetic accounts
Authentication	Package-declared BETTER_AUTH_SECRET and application auth code	Verify the secret reaches the auth system; test login, logout and cross-account isolation
Optional SMTP	Notification helper and global settings	Confirm credential handling, provider requirements and actual receipt

Treat notifications as a separate delivery test

Begin without real SMTP credentials. The pinned code reads notification settings from D1 and schedules email work after storage. It catches notification errors rather than making delivery a condition of submission success. A dashboard record and a successful form response are therefore different evidence from an email arriving.

If notifications are required, use an operator-authorized test inbox, review who can access settings and backups, and test the chosen SMTP provider on the deployed runtime. The README's planned Resend path does not prove Resend is the current implementation. Provider charges and domain costs belong in a separate budget.

Estimate quotas from measured operations

Workers Free currently allows 100,000 requests per day across the account and 10 ms CPU per invocation. D1 Free allows 5 million rows read per day, 100,000 rows written per day and 5 GB total storage. Dashboard reads, authentication, migrations and notifications can use resources beyond the single submission insert. Measure a representative workload before translating request limits into a submission capacity claim.

Complete a reversible form pilot

1. Use separate resources, synthetic accounts and a non-critical test form.
2. Test HTML and JSON submissions, missing form IDs, invalid payloads and redirects.
3. Verify dashboard access, exports and cross-account isolation; assess bot traffic and abuse handling before public exposure.
4. If required, test actual notification receipt and an observable failure case.
5. Keep the old form endpoint available for rollback while checking the new route.

Compare the FormZero workflow, Formspree plans and other alternatives before cutover. Source review and the author's real dashboard image support evaluation; they do not certify a fresh deployment or production readiness.