Skip to main contentGet your mobile app iPhone Duo ready

EngineeringCodeCross TeamUpdated

Harden a FlutterFlow MVP before paid ads (2026)

Harden a FlutterFlow MVP before store traffic: write owner rules per record, keep payment secrets off the client, and prove sign-in.

Engineering

13 min

Rules
Deployed

Not canvas only

Secrets
Private API

Not Custom Action

Ads
After

Store or domain first

Harden a FlutterFlow MVP before paid traffic by deploying Firestore rules that name the owner of each row — not “Authenticated Users” for every signed-in stranger. Use a private Environment Value only inside a private API call. Keep Stripe secret keys out of Custom Actions and public API headers. Point ads at one origin you proved: the store binary or the custom domain, not a leftover flutterflow.app Test Mode tab. You may pause an Apple phased release for up to 30 days, with no limit on the number of pauses. Halt a Play staged rollout. Then ship a new higher build. Then buy ads.

CodeCross is a Delaware-registered company with a postal address in Austin, Texas. Product and engineering are in Pakistan. Other builder guides live on vibe coding. If you want us, start at Austin app development company. If nobody can name a store recovery procedure tonight, book a conversation.

Money leaves when the campaign still lists a flutterflow.app Test Mode tab, a Firestore collection left on Everyone, or a Stripe secret pasted into a Custom Action. FlutterFlow pricing puts Basic at $39 a month. That fee buys code download, an APK, and one-click store deploy. It does not buy a closed row. Firestore rules start a new collection at Create Everyone and Read Everyone. A setting you never Deployed is not live.

This essay is the week before you buy ads, while the app still comes from FlutterFlow. Close the risky doors while the builder still writes the tree. It expands Harden a FlutterFlow MVP before strangers download. The question here: can you buy a click while FlutterFlow still builds the app?

Stay in FlutterFlow. Close the doors first.

FlutterFlow is paid to finish a screen tonight. Paid traffic is paid to send strangers at a public binary. Those jobs collide. A green Test Mode run can still leave Everyone on a notes collection, a secret in a Custom Action, or a flutterflow.app name as the share link.

FlutterFlow’s own docs draw the line. Development Environments start every project on Production. Extra environments are a paid add-on. Switching an environment only changes the Firebase project or the Environment Values. It does not close a row. You still pick who may read a document. You still pick which URL ads will hit.

Web publishing will put a live site on your-project-id-1234.flutterflow.app for free. Play deploy and App Store deploy are a different door. You do not need a new builder this week. You do need to name who holds Firebase, who holds store seats, and which URL ads will hit.

Fix the biggest risk first

Start with the worst risk, not a prettier home screen. A stranger who dumps every order row, burns a paid API from a public header, or copies a live Stripe secret will cost more than a new landing page. Put those risks in order, worst first. Close the worst one tonight.

Worst risk first — close the top FlutterFlow gap

  1. Off switch and watch

    An Apple pause of up to 30 days, with no limit on the number of pauses, or a Play halt you have run once, plus a new higher build — not a one-hour snapshot on the Free plan.

  2. Code download mistaken for the train

    A zip or a flutterflow branch push is not a protected main. The next push overwrites that branch.

  3. Store review still unfinished

    No demo account. Placeholder copy. Data safety that does not match the SDKs in the binary.

  4. Test Mode vs the origin you own

    Ads, mail, or Auth that still print a flutterflow.app Test Mode tab or a leftover project-id subdomain.

  5. Secrets in Custom Actions and public APIs

    A Stripe secret in Dart. A private Environment Value used outside a private API call. A key in Info.plist.

  6. Open Firestore or Supabase rows

    Everyone or Authenticated Users on a private collection. RLS left off. A dated test-mode rule still live.

Ads multiply whatever is already open. A Test Mode pass does not seal rules. A flutterflow.app tab is not a campaign host. “We will revert a snapshot” is not a store recovery procedure.

Put six names on paper: who owns Firebase or Supabase, who proves login on the advertised binary, who can click Deploy, who watches spend and errors, who owns the public origin, and who can pause a phased release, halt a Play rollout, or ship a new higher build tonight. Two red names plus a campaign date means spend stays off. If you cannot name a human for each layer, book.

Firestore rules: only deployed rules lock rows

Rows are the first thing ads will leak. FlutterFlow is good at drawing a list. It is weaker at naming who may read a row after that list. Firestore rules are the server list that says who may create, read, write, or delete a document. A hide on a widget is paint. It is not that list.

A new collection starts open. Create is Everyone. Read is Everyone. Write is No One. Delete is No One. FlutterFlow marks private collections with Has Private Data and warns you. A notes collection that lets everyone read every note is the example they use.

Authenticated Users is the next trap. That setting means any person who signed in — Email, Google, or another method — can create, read, write, and delete. It does not mean “only the person who made this row.” Firebase’s insecure-rules guide says the same thing in other words. A rule that only checks request.auth.uid != null lets any logged-in user into the whole set. If one of your rules is that wide, confirm you really want every signed-in stranger in.

Tagged Users is the owner lock FlutterFlow already ships. The document must hold a user reference or a user id. Users Collection is only for a users table, and it matches the auth id to the document id. Use those when the row is personal. Do not hide an admin page and call that a gate.

A setting you never Deployed is not live. You must hit Deploy every time you change a rule. Before you publish, the same page says to remove leftover test-mode lines such as allow read, write: if request.time < timestamp.date(2024, 5, 31); and exit Test mode. OWASP Top 10 2025 still ranks broken access control first. They found some form of it in 100% of apps they tested. One example is an admin check that lives only in the front end.

Supabase: public anon key, row rules guard data

If the project uses Supabase — a hosted Postgres database — day one is just as open. Supabase setup lets you uncheck Enable Row Level Security so you can move fast. The same note says to turn Row Level Security back on and write a policy before you deploy. Row Level Security is the set of database rules that decide who may read each row. Row level security is a lock inside the database that checks each row.

The anon key you paste into FlutterFlow is meant to sit in the app. It is a project id, not a vault. Supabase’s RLS guide is blunt. A table in an exposed schema without RLS is open to any role with a grant. Adding a policy does not revoke default anon and authenticated grants. Prove both. Development Environments want a separate Supabase project per environment, with SupabaseAPIURL and SupabaseAnonKey as Environment Values.

Secrets: a Custom Action ships in the phone.

Keys are the next thing ads will leak. A Custom Action is a Dart snippet you paste into FlutterFlow. That snippet ships in the APK and the IPA. Anyone can unpack it. Founders then drop Stripe secrets and webhook tokens into the same action. Rotate anything that sat in plain text.

An Environment Value is a named setting that can change between test and live. You can mark one private. Development Environments say a private value is not put in the compiled app. The catch: the only way to use it is as a variable on a private API call. A private API call is a call FlutterFlow sends through a Firebase Cloud Function so the phone never sees the secret. Drop that value into a Custom Action and it is back in the phone.

The Cloud Function file is a second leak. The same page says a private value used in a private API call may appear in the generated function. Review those files before a GitHub push. API calls repeat the rule. Turn on Make Private. Save. Deploy APIs. You can Require Authentication so only a signed-in Firebase user can hit that function. A public API call with Authorization: Bearer YOUR_TOKEN in the header still sends the token from the phone.

Configuration files are public too. Anything in Info.plist or AndroidManifest.xml ships in the distributed app. FlutterFlow says do not assume a key in Info.plist is hidden. For keys you must include — maps, for example — use a private Environment Value and watch spend. Stripe’s API keys guide is the split. A publishable key starts with pk_ and can sit in the app. A secret key starts with sk_ and cannot. Only publishable keys are safe outside the backend. If the only place the secret still lives is a Custom Action you are afraid to open, book.

FlutterFlow gives you several addresses that look live. They are not the same job.

Three live origins — only one is the ad target

  1. 01 · Test Mode / Run

    Lab

    In-builder preview. Dies when the session ends. Not the ad.

  2. 02 · flutterflow.app

    Share door

    A project-id subdomain or a custom subdomain. Old names can move.

  3. 03 · Store or your domain

    Ads

    The binary on Play or App Store, or the custom domain you proved.

Test Mode is the lab. A flutterflow.app publish is a share door. The store binary or the custom domain is the name you own. Mixing them is how the builder becomes the company.

Web publishing is honest about the share door. The default URL is your-project-id-1234.flutterflow.app. A custom mywebapp.flutterflow.app name counts against the cap — two on Free, twenty on paid. Change the name and it only applies on the next publish. Old addresses can stop working and go to someone else. If you use Firebase Auth, add that subdomain as an authorized domain or social and phone sign-in will fail.

The store binary is a different origin. Play deploy needs a service-account JSON. The first release still needs you to download the AAB — the Android App Bundle Play wants — and upload it to Internal testing. App Store deploy needs the Apple ID, a Team Key, the Issuer ID, and the Key ID. Deploy sends a build. The mail that follows is “a build arrived,” not review.

Google Ads destination mismatch rejects an ad when the display URL domain does not match the final URL. Do not advertise your domain if clicks still land on a flutterflow.app Test Mode tab. Do not advertise a store listing if the final URL is still the project-id subdomain. Point ads, mail, and in-app shares at one name you proved.

Store review: Test Mode is not App Completeness

Stores review the binary, not the builder canvas. Apple’s App Review Guidelines put App Completeness at 2.1. Submissions should be final. Placeholder text and empty websites should be gone. Test on a device. If the app has login, give a demo account and turn the back-end on. Incomplete or crashing binaries are rejected.

Play Data safety is the Android twin. Every developer with an app on Play — including closed, open, or production testing — must fill the form and say what the app and its SDKs collect. Apps only on an internal testing track are exempt from the listing section. That exemption is not a reason to skip the form before you promote. Internal testing is also the first Play track FlutterFlow asks you to use.

A FlutterFlow watermark or a page that still says lorem is placeholder content. Paid plans can hide the watermark. Hiding the badge does not finish 2.1. Name the reviewer login. Prove the money path on the advertised binary.

Code export is a copy that gets overwritten

Basic can download source and an APK. Growth adds GitHub. Push to GitHub always writes a branch named flutterflow. The next push overwrites it. Keep custom Dart on a develop branch. Merge with a pull request. Deploy from main. A zip on a laptop is not that train.

A download is useful. It is not the wrap. If the live binary still comes from one-click deploy and nobody has paused an Apple phased release for up to 30 days, with no limit on the number of pauses, halted a Play rollout, or shipped a new higher build, ads wait.

Off switch: a snapshot is not rollback

A bad release needs an off switch you can name tonight. Reverting a snapshot takes the builder back. It does not take the store listing back by itself.

Saving and versioning splits three drawers. Project Versions are deprecated. Commits are manual saves on a branch. Snapshots are automatic saves as you edit. Free keeps one hour. Basic keeps one day. Growth keeps three days. Business keeps seven. Revert puts the canvas back. It does not republish Play, App Store, or the custom domain.

Web publishing has View Full History for prior web deploys. Unpublish takes the site down. Name the store owner. Prove you can pause an Apple phased release for up to 30 days, with no limit on the number of pauses. Halt a Play staged rollout, or ship a new higher build. “We will revert the snapshot” waits on one plan’s retention window. It is repair. It is not an off switch.

Watching: errors, spend, and silent Test Mode

Watching without an off switch is half a lock. Before ads, name three signals: sign-in failures on the advertised binary, crashes after a Deploy, and paid-API or Firebase spend. Google Analytics needs Firebase first. Enable it in the Firebase console, then toggle it in FlutterFlow. You can log page load as screen_view, button taps, and auth events. Firebase may take up to 24 hours to show events. Name who watches the first day.

Web size is the other watch. Web publishing says Import Emoji Library raises the size of the site. CanvasKit can look sharper and then trip CORS — a browser rule that blocks images from another host. Prove the advertised web origin on a cold load before you buy a click that lands there.

Score — mark pretend vs proof

Firestore / RLS

  • Costume

    Everyone / RLS off

  • Started

    Rules drafted

  • Proof

    Deployed + stranger denied

Secrets

  • Costume

    Key in Custom Action

  • Started

    Moved to private env

  • Proof

    Private API + rotated

Public name

  • Costume

    Ads list Test Mode

  • Started

    Domain or store live

  • Proof

    Ads + Auth match

Store forms

  • Costume

    Placeholder copy

  • Started

    Listing drafted

  • Proof

    Demo account + Data safety

Export / Git

  • Costume

    Zip on one laptop

  • Started

    flutterflow branch pushed

  • Proof

    develop + review

Off / watch

  • Costume

    “Revert snapshot”

  • Started

    Prior version named

  • Proof

    Prior store version restored

A green Test Mode run is not proof. A flutterflow.app publish is still a share door — and that is fine for this page if public ads no longer print it.

Score each row against the advertised binary or the custom domain, not Test Mode. One red cell — Everyone on a private collection, a secret in a Custom Action, a flutterflow.app ad, a missing demo account, a silent spend meter, or a rollback you have never hit — is enough to hold the spend. Move that cell. Do not greenwash the matrix.

The week before you buy ads

Treat the next seven days as operator time on the FlutterFlow project, not a slogan. Inventory who can Deploy, who holds Firebase, and which origin ads will hit. Then seal. Then prove.

Harden in FlutterFlow — seven operator steps

  1. 01 →

    Inventory the public path

    Who can Deploy. Which origin ads will hit. Where secrets still live. Test Mode vs store named.

  2. 02 →

    Seal rows

    Tagged Users or owner rules. Deploy Firestore. Turn RLS on. Remove dated test-mode lines.

  3. 03 →

    Seal secrets

    Drain Custom Actions and public API headers. Rotate. Private Environment Values only on private API calls.

  4. 04 →

    Fix the URL map

    No flutterflow.app in ads. Auth authorized domains match the custom domain or store host.

  5. 05 →

    Finish store forms

    Demo account. Live backend. Data safety that matches the SDKs.

  6. 06 →

    Name watch and spend

    Who sees sign-in fails, crashes after Deploy, and the Firebase or Stripe bill.

  7. 07

    Restore + watch, then buy

    Pause an Apple phased release for up to 30 days, with no limit on the number of pauses. Halt a Play staged rollout once. Then spend.

Do not buy ads in step one. Do not start the builder exit while a step is still open.

Day 0 is inventory. Write five facts. The public origin. The person who can Deploy. Whether a live key still sits in a Custom Action. Whether a stranger can read another user’s row. Whether you can pause a phased release, halt a Play rollout, or ship a new higher build tonight. If the public origin is already taking leads and you cannot name those five, stop. You do not need a new screen. You need this list.

Seal rows first. Then secrets. Then the URL. Bots do not wait for a nicer landing page. If a public collection exists, prove it refuses the read when the stranger is not the owner. Then strip leftover flutterflow.app shares. Then pause an Apple phased release for up to 30 days, with no limit on the number of pauses, or halt a Play staged rollout in a safe window so you know the off switch works. Ship a new higher build if the bad binary already shipped. Then spend. This page is the week you run that list.

Example operator days before a FlutterFlow ad buy

days

Unpriced Ads on a red list

3–5 wks

Campaign live. Doors still open. Cleanup later.

08162432Example calendar only, not a quote or a promiseDay 0–1Inventory + seats1 dayDays 1–3Rules + RLS1–3 daysDays 2–5Secrets + private APIs2–5 daysDays 3–7Origin + rollback3–7 daysUnpricedAds on a red list3–5 wks

Example calendar only, not a quote or a promise

Example operator days for planning. These are not measured project results or a quote.

Reserve these days before paid traffic. Overlap is allowed. Skipping inventory to “buy a small test” is how Test Mode becomes the company.

Treat those bars as calendar you reserve, not a vendor promise. If you cannot name the advertised origin, who can Deploy, and who can pause a phased release or halt a staged rollout, the later days will thrash.

Stay in FlutterFlow, or leave. Not both.

Pick one job for this week. Stay keeps FlutterFlow as the studio, with Deployed owner rules, private API secrets, a store or custom domain as the origin, and a store recovery procedure already proved. Leave opens get-off FlutterFlow and stops buying clicks that need that builder window.

Stay or leave — pick one job

Stay (this page)

  1. Studio still in FlutterFlow

    The builder still plans. You accept the canvas and the plan bill.

  2. Doors closed on that path

    Rules, secrets, origin, store forms, spend, rollback.

  3. Ads wait for green

    A red layer pauses spend, not the product.

If the public origin must leave a builder-only ship path, stop this page and open the exit. If ads are close and the origin can stay on a store or domain you control, finish these doors first. ↓

Leave (other pages)

  1. Process leaves the builder loop

    Org Git plus a host you run. See get-off and migrate.

  2. A flutterflow branch is not the exit

    The next push still overwrites that branch until your host serves the live app.

  3. Kill the builder-only ship

    App still answers when the FlutterFlow tab is closed.

This article is the left column. Get off FlutterFlow and migrate from FlutterFlow are the right. To-production is the first host you run yourself. Do not paste them here.

If you only need a leave date, use when to leave an AI builder. A door that will not close belongs on rewrite vs harden. A personal Git remote belongs on GitHub handoff. A builder-only ship path belongs on CI/CD after an AI builder. Other builders have their own pages on the vibe coding hub.

Check loop — prove, then decide

  1. 01

    Prove the door

    One layer. One test on the advertised binary or domain.

  2. 02

    Close it

    Deploy rules, rotate, redirect, add identity checks, or pause a phased release, halt a Play rollout, or ship a new higher build.

  3. 03

    Re-check ads list

    If a layer is still red, spend stays off.

  4. 04 · loops

    Stay or leave

    Closed doors can stay. An open builder-only row opens the exit.

Do not buy ads inside the loop. Do not leave FlutterFlow inside the loop. Exit the loop when the doors are closed or the live path must move.

When the list turns red

Leave the product up. Kill the campaign if any of these are still true:

  • A private collection still allows Everyone or Authenticated Users, or a dated test-mode rule is still live.
  • A live key still exists in a Custom Action, a public API header, Info.plist, or a committed Cloud Function file.
  • Ads, mail, or the sitemap still list Test Mode or a leftover flutterflow.app name.
  • Store review still lacks a demo account, a live backend, or a Data safety form that matches the SDKs.
  • The off switch is “we will revert a snapshot,” or you have never paused a phased release, halted a Play rollout, or shipped a new higher build.
  • Two people disagree about who can Deploy or who holds Firebase.

A single red door is a miss, even if Test Mode is green. Stop adding screens until rules, secrets, the origin, and the rollback are honest. Finish those doors. Then spend — or open get-off FlutterFlow if the builder is still the company.

Bring a second pair of hands when the founder cannot rotate a key without pasting it back into a Custom Action, when a stranger can still find another user’s row, or when nobody has paused a phased release, halted a Play rollout, or shipped a new higher build before the date. Hire for rules, secret locks, Deploy seats, and the URL map — not a prettier home screen. Austin app development company is the company brief. Austin mobile app development if the next door is a store binary that still points at this FlutterFlow-built URL.

Next steps

Walk the gates in order. Inventory the public origin, the person who can Deploy, and who holds Firebase. Drain Custom Actions into private API calls. Rotate anything that appeared there. Deploy owner rules. Turn RLS on. Strip leftover flutterflow.app shares. Prove a stranger test on the live binary. You may pause an Apple phased release for up to 30 days, with no limit on the number of pauses. Halt a Play staged rollout. Then ship a new higher build. Then buy the click — or open the exit if the process itself must leave FlutterFlow.

CodeCross LLC is a Delaware-registered company with an Austin postal address. Product and engineering are in Pakistan. The FlutterFlow week we run is Deployed owner rules, private API secrets, a store or domain you own, store forms that match the binary, then a store recovery procedure you have already hit. The Austin app development company page is the company brief. Austin mobile app development is the store-binary door if a signed build still points at this FlutterFlow-built URL. Our apps are on the App Store and Google Play. When the off switch, secrets, or the advertised origin is still open, book a conversation.

FlutterFlow is allowed to stay the studio. Ads are not allowed to treat a Test Mode pass, an Everyone rule, or an unreviewed Custom Action secret as that proof. Show an Apple pause of up to 30 days, with no limit on the number of pauses, or a Play halt you can run, plus a new higher build, rows that refuse a stranger who is not the owner, and keys that never lived in Dart. Most teams never need a second codebase once those three exist.

Next steps: migrate from FlutterFlow, or plan a move away from FlutterFlow.

FAQ

Does “Authenticated Users” stop one signed-in stranger from reading another user’s order?

No. Firestore rules say Authenticated Users lets any logged-in person create, read, write, and delete. It does not check who owns the row. Firebase’s insecure-rules guide flags request.auth.uid != null as open to every signed-in user. Use Tagged Users with a user id field, or write a custom rule in the Firebase console. Deploy after you change it. A hide on a widget is not that proof.

If I mark an Environment Value private, can a Custom Action still ship it in the APK?

Yes, if you use it there. Development Environments keep a private value out of environment.json and out of FFDevEnvironmentValues. The only supported use is as a variable on a private API call, which rides a Cloud Function. A Custom Action is Dart in the phone. Put the value there and it ships. Review generated Cloud Function files before a GitHub push. Rotate anything that sat in Dart.

Does downloading source or pushing the flutterflow branch finish the wrap?

No. Push to GitHub always writes the flutterflow branch and overwrites it on the next push. A Basic code download is a zip of tonight’s tree. Neither one deploys Firestore rules, rotates a Custom Action secret, or puts an older store binary back. Keep custom work on develop. Merge with a pull request. Deploy from main. A laptop zip is not that train.

If I revert a FlutterFlow snapshot, does the store listing go back?

No. Saving and versioning says snapshots are automatic builder saves. Free keeps one hour. Basic keeps one day. Growth keeps three days. Business keeps seven. Revert puts the canvas back. It does not republish Play, App Store, or the custom domain. Apple will not put an older App Store version back. You may pause the release for up to 30 days, with no limit on the number of pauses. Then ship a new higher version. On Play, halt a staged rollout so fewer people get the bad build. People who already updated stay on it. Hit a feature flag if you have one.

Does a green Test Mode run count if Firestore still has a dated open-access rule?

No. Firestore rules say you must Deploy, and you must remove leftover lines such as allow read, write: if request.time < timestamp.date(2024, 5, 31); before you publish. Test Mode can look fine while that line is still live. A stranger with the project id can still read or write until the dated rule expires — or forever, if someone later set if true. Prove the stranger test on the advertised binary after Deploy.

Can I advertise a flutterflow.app subdomain if the store listing uses a different package?

No. Web publishing puts the default site on your-project-id-1234.flutterflow.app. Old custom subdomains can stop working and go to someone else. Google Ads destination mismatch rejects an ad when the display URL domain does not match the final URL. If the store listing is the product, the ad must open that listing or the domain that serves the same app. Do not mix the share door and the store door.

Book a call

Thirty minutes with a senior teammate — honest next steps.

Want a second pair of eyes on your app?

Tell us what you built and where it's stuck. We'll tell you what it takes to make it ready for real users.

Prefer writing? Send project details on the contact page.