# Vibe Coding Production Checklist

50 checks for apps built with Cursor, Claude Code, Codex, Lovable, Bolt, Replit or v0. Tick a check only after you have evidence. Mark inapplicable checks with a written reason. A completion score measures your recorded checks, not the probability that an app is safe.

Source: https://lab.anonyfox.com/vibe-code-production-checklist/
Version: 1.0 · build dated 2026-11-10. Free to share unchanged with attribution.

Start with server-side access control, client-visible secrets, payment replay, concurrent writes and recovery. Use an isolated environment for behavioral checks; the supplied scanner does not run these tests.

## Auth & access

- [ ] **Every sensitive route authorizes on the server.** Verify: Trace registered middleware; call the route as a normal user and while logged out.
- [ ] **Tenant ownership covers reads, writes and deletes.** Verify: Use two tenant accounts; cross-tenant requests must fail without changing state.
- [ ] **Supabase RLS or Firebase rules restrict each user.** Verify: Inspect deployed rules and test anonymous, owner and other-user access.
- [ ] **Sessions expire and reset links can be used only once.** Verify: Test expiry, logout and a repeated reset link against an isolated environment.
- [ ] **Admin roles come from verified identity.** Verify: Change a body/header role field; it must not grant privileges.

## Secrets & bundles

- [ ] **Privileged keys stay out of browser code.** Verify: Inspect the built JavaScript for secret references; publishable keys are different.
- [ ] **NEXT_PUBLIC_ and VITE_ names contain only public values.** Verify: Review each public variable and its value class without copying credentials.
- [ ] **Environment files stay out of version control.** Verify: Check ignore rules and tracked filenames; review history separately if exposure is suspected.
- [ ] **Exposed credentials have been revoked.** Verify: Confirm rotation with the provider and remove every client or logged copy.

## Validation & injection

- [ ] **Server schemas reject invalid input.** Verify: Send wrong types, unknown fields and oversized values; expect controlled rejection.
- [ ] **Queries bind values as parameters.** Verify: Trace untrusted input into SQL; check that values cannot change query structure.
- [ ] **HTML rendering uses an appropriate safe output context.** Verify: Trace raw HTML sinks to trusted or maintained sanitization paths.
- [ ] **Uploads and fetched URLs have boundaries.** Verify: Check content/size limits, storage permissions and private-network URL rejection.

## Payments & webhooks

- [ ] **Webhook signatures are verified before effects.** Verify: Alter a signed test payload; verify rejection before fulfillment writes.
- [ ] **Duplicate events cannot duplicate fulfillment.** Verify: Deliver the same event concurrently and after restart; expect one business effect.
- [ ] **Outbound payment retries use durable idempotency keys.** Verify: Retry the same operation after a simulated connection loss; expect one charge.
- [ ] **Amounts and entitlements come from trusted server state.** Verify: Change client price/customer fields; they must not control fulfillment.
- [ ] **Refunds and failed renewals update access.** Verify: Run test-mode lifecycle events and check persisted entitlement state.

## Data integrity

- [ ] **Multi-step writes commit atomically.** Verify: Fail each step in staging; earlier dependent writes must roll back.
- [ ] **Concurrent requests preserve business invariants.** Verify: Race two stock, balance or quota updates; neither may overdraw the resource.
- [ ] **Uniqueness and relationships are database constraints.** Verify: Attempt duplicate and orphan writes directly in an isolated database.
- [ ] **External effects have an outbox or durable recovery path.** Verify: Crash between state change and dispatch; restart must converge without duplication.

## Queries & performance

- [ ] **Lists have pagination and bounded results.** Verify: Seed representative data; verify stable bounds and cursor behavior.
- [ ] **Loops avoid one database call per entity.** Verify: Compare query count for 1, 10 and 100 entities; inspect batching.
- [ ] **Hot filters and joins have measured query plans.** Verify: Review EXPLAIN plans on staging data near the expected production volume.
- [ ] **Expensive work has bounded concurrency.** Verify: Submit a burst and observe queue depth, worker count and response budgets.

## Resilience

- [ ] **External calls have per-attempt and total deadlines.** Verify: Simulate a dependency that never replies; requests must release workers on time.
- [ ] **Retries are bounded with backoff and jitter.** Verify: Force transient errors; count attempts and check delays against the request budget.
- [ ] **Repeated upstream failure is contained.** Verify: Verify a circuit breaker or equivalent concurrency cap prevents worker exhaustion.
- [ ] **Background jobs survive process restarts.** Verify: Kill a worker around acknowledgment; work must complete once or enter a failed-job queue.

## Errors & observability

- [ ] **Failed writes reach the caller as failures.** Verify: Break a staging write; the UI must show failure and permit a safe retry.
- [ ] **Logs redact credentials and personal records.** Verify: Trigger representative failures and inspect redacted structured logs.
- [ ] **Critical errors reach a responsible person.** Verify: Send a synthetic alert and confirm delivery and acknowledgment.
- [ ] **Health checks distinguish alive from ready.** Verify: Simulate a failed dependency; readiness should fail without exposing internals.

## Deploy & recovery

- [ ] **A clean database can be rebuilt from migrations.** Verify: Apply migrations from zero in staging and exercise a core workflow.
- [ ] **A recent backup has been restored successfully.** Verify: Restore to an isolated environment; record recovered data and elapsed time.
- [ ] **Staging and production use separate write credentials.** Verify: Inspect environment bindings and payment modes without exposing values.
- [ ] **Rollback works with the changed schema.** Verify: Rehearse the documented rollback after a representative migration.

## Rate limits & AI abuse

- [ ] **Login and reset endpoints have enforced limits.** Verify: Burst through the real ingress; verify 429 responses and per-user fairness.
- [ ] **Paid AI endpoints reserve and cap spending.** Verify: Race requests near the budget; total reservations/spend must stay within the cap.
- [ ] **Uploads, exports and invitations have abuse limits.** Verify: Exercise repeated expensive requests; verify bounded work and recovery.

## Dependencies & supply chain

- [ ] **Dependencies match intended maintained upstream projects.** Verify: Review names, lockfile provenance and install hooks without executing them.
- [ ] **Current vulnerability advisories are reviewed.** Verify: Run an authorized advisory scan separately; evaluate reachability and supported fixes.
- [ ] **CI actions and credentials have narrow trust boundaries.** Verify: Inspect immutable action refs, fork triggers and job permissions.

## Tests that matter

- [ ] **Auth and money paths have behavioral regression tests.** Verify: Check assertions against real persistence and authorization boundaries.
- [ ] **Failure and concurrency tests can catch the named failure.** Verify: Inspect duplicate-event, failed-write and race test scenarios and expected state.
- [ ] **CI actually runs the full required suite.** Verify: Check focused/skipped tests and ignored failure exits in the workflow.

## AI-specific review

- [ ] **SDK methods exist in the installed version.** Verify: Check supported signatures/types or official source for critical integration calls.
- [ ] **Critical paths contain working implementations.** Verify: Trace TODOs, stub returns and mock success states reached by real users.
- [ ] **One business rule has one clear source of truth.** Verify: Read oversized modules and duplicated workflows; compare their invariants.

## Decide what to fix first

For each actual finding, estimate consequence severity × likelihood × blast radius on scales of 1–4. Products ≥36 are Critical; 18–35 High; 6–17 Medium; 1–5 Low. Explain assumptions. These are prioritization estimates, not probabilities or CVSS. A checklist cannot decide whether an unverified issue exists.

The interactive page orders unchecked items by default impact estimates. It does not inspect your code. Adjust priorities in the actual review to your application and deployment.
