App Builder FAQ
Answers to common setup and operating questions about App Builder.
For what the app is and how to set it up from scratch, see the
App Builder Overview.
Setup
Does App Builder need to be enabled or licensed before anyone can use it?
There is nothing to license. App Builder comes with every edition. On a plan that includes it (every edition, Platform Free and above) it is on unless an administrator turns it off in Admin → Apps → All Apps; otherwise an administrator turns it on there. Building and trying apps is free, and there is no per-employee charge. Where mini app packs are switched on (our shared cloud), each published mini app is counted in a pack that follows your live apps; custom apps are not counted. Turning App Builder on shows no built app to anyone: every app starts unpublished until an administrator or App Builder admin publishes it. AI tokens the builder spends stay on the shared platform AI-credits meter and are not billed separately.
Who can build apps, and who can only monitor?
Three tiers. A global administrator can do everything: create and edit mini apps, build, edit and deploy custom apps, manage API keys, and change settings. The composer tier — governed by the Who can create mini apps setting on App Builder → Settings (the Who can create mini apps card) — can compose, edit, and start mini apps from templates, and sees the mini apps they created on Your apps: the default admits everyone with the Manager role (plus admins and app-admins). A team lead with the Member role is not in it until you choose Managers and admins, plus anyone with direct reports; you can also widen it to everyone or narrow it to app-admins only. A delegated app-admin — granted from Admin → Apps → All Apps — can monitor Your apps, publish and unpublish apps there, remove a mini app or retire a custom app, browse Templates, view Analytics, and retry loading a failed app. Code builds and Settings stay global-admin everywhere. Anyone else who built an app asks for it with Ask an admin to publish on the app’s own page (also on its row in Your apps, its Edit with AI page and on the phone), where they can also see that they asked and withdraw the request.
Can my managers create mini apps without me granting each one access?
Yes — that is the default for everyone with the Manager role. Who can create mini apps (App Builder → Settings) ships set to Anyone with the Manager role or above, so they can open App Builder’s home, see the mini-app and template cards, and compose — with no per-person grant. People who run a team but have the Member role are not included: choose Managers and admins, plus anyone with direct reports to add them without leaving anyone out. Set it to Everyone with access to the app and any member can (the office manager who needs a visitor log builds it without asking); set it to App admins only to require an explicit grant again. Whoever composes also approves the app’s capability grants on its card, so widen with that in mind.
I granted someone app-admin access but they can’t build a custom app — is that a bug?
No. Building a custom app deploys code and Settings governs unattended AI spend — both require the global-admin tier. A delegated app-admin sees Home, Your apps, Templates, and Analytics; they can publish and unpublish apps from Your apps, remove a mini app or retire a custom app there, and compose mini apps (the composer tier admits app-admins). The surfaces match the actions each tier can take, so nothing they can click will bounce them.
Where did the Build, Compose, and Builds tabs go?
They were removed from the left rail on purpose. Tabs are now places — Home, Your apps, Templates, Analytics, Settings — and the verbs live in the Create fork on the Home page: Mini app, Custom app, and Start from a template. Build history moved onto Your apps and the Recent builds card on Home. Every old URL still works; only the tabs changed.
Mini apps vs. custom apps
Should I create a mini app or a custom app?
Start from what the thing is, not from how it should be built. If you are tracking, counting, or logging records — supplies on a shelf, visitors at a desk, equipment on loan, vehicle inspections — that is a mini app: no code, reviewed on a plain-language approval card, and running the same day. If you need custom screens, unusual logic, or an integration with an outside system, that is a custom app and AI writes real code for it.
You do not have to get this right up front. The create fork asks in those terms rather than by technology, and the two paths cross-recommend each other — start describing a tracker in the code builder and it will point you at the faster path.
What is actually created when I approve a mini app?
The app and what it owns: its lists with your fields, the saved views you approved, its rules, and — if the plan included one — a live board. They are registered as a marketplace app with its own slug, icon, launcher entry, and permissions, running at /apps/<slug>. No code is generated and nothing is deployed — a mini app is lists, views and rules the app itself owns, so its records, views and rules live inside the app. That is why a mini app is live in minutes.
Do mini apps show up in the public app directory?
No. Mini apps are private to the tenant that created them. They never appear in the public MangoApps directory or in another business’s marketplace, and there is nothing to submit — listing an app for other businesses exists only for custom apps. Publishing a mini app (next question) makes it live inside your own company, nowhere else.
Are my mini apps visible to employees straight away?
No. Every app you build — mini or custom — starts unpublished: only the person who built it, your administrators, App Builder admins and anyone given a role on the app can open it, so you can try it with real records before anyone else sees it. To take it live, an administrator or App Builder app-admin presses Publish on its row in App Builder → Your apps, which shows it to everyone; to limit it to some groups, add a visibility rule from Admin → Apps → All Apps instead. If you built it but cannot publish, press Ask an admin to publish on the app’s own page or its row in Your apps — it then says that you asked, offers Withdraw request, and tells you if an admin declines. Unpublish on its Your apps row takes a live app back to the person who built it, your administrators, App Builder admins and anyone given a role on the app — except that groups it is shared with keep it until their visibility rule is removed in Admin → Apps → All Apps.
Do I need Power Tables and Live Boards licensed separately?
No, and you do not need to turn them on. A mini app’s records belong to the mini app itself, so it works with neither app enabled, and approving one switches nothing else on for the business.
What a mini app can and cannot do
What can a mini app’s rules actually do?
The always-on palette is four actions: update a field on the record, notify a person, open a task, and send a scheduled digest of what a saved view matches. Recipients come from a fixed vocabulary — the record’s filer, the app’s creator, a person-field on the record (an assignee), that person’s manager, a role, or a team — never a free list.
Four more powers exist only as capability grants the approval card disclosed and you approved: send email to an email-address field stored on the record (one address, rate-capped), request approval from a named approver before something proceeds, public capture (a public form at a tokened link), and kiosk capture (a shared-tablet kiosk where a badge tap identifies the employee — see the kiosk question below). You can revoke any of them later from the edit loop, and the revocation card names everything that stops working. The AI never drafts, at any tier, a webhook, a rule that deletes or archives rows, an AI step, or an approval nested inside an approval. An app’s admins can still add the first three by hand on the app’s Rules page.
Can I get an alert 30 days before something expires?
Yes. A mini app’s triggers are a record was created, a field changed, and on a schedule — a scheduled rule binds to a saved view and sends a digest of everything the view currently matches. So “every Monday, message me what expires in the next 30 days” is one rule: a view filtered to the expiry window plus a scheduled digest to you. The digest is one summary message, not one message per row, so a big backlog cannot flood anyone.
Can one mini app cover two linked tables?
Yes — up to three. Describe owners, pets, and stays and the drafter composes three linked tables, connecting them with reference fields (a stay points at its pet, a pet at its owner) and giving every table its own page. Describe more entities than fit and it folds the extras into fields on an existing table — and the approval card says exactly where the fourth list went, so you approve what actually happened. A mini app can also hold a reference to another mini app’s records, which is the platform’s answer for shapes bigger than one app.
Can the edit loop rename or remove things, or only add?
Both directions. Adding: fields, views, rules, pages, a linked list, a board, a quick-update form, a public form. Changing and removing: rename a field, remove a field, remove a page, edit a workflow field’s allowed status moves, and revoke a capability. Removals refuse honestly instead of half-working — you cannot remove a list’s only page, and an app always keeps at least one page. An app’s admins can also add, rename or archive fields on its Fields page and bring records in from its Import data page; splitting a list or moving records between lists is not something any page does.
Can people without OneMango accounts submit records to a mini app?
Yes, if the app holds the public capture grant. The approval card discloses it in plain words (“anyone with the link can submit — no sign-in”), and once approved the app serves a public form at a tokened link you’ll find in the app’s settings — post it at the gate, print it as a QR code, or email it to suppliers. Submissions are honeypot-protected and rate-capped per form per hour. Fields the app’s own rules set (a governed Status, say) are never offered on the public form — outsiders state facts, the app runs the workflow. Public submissions land as ordinary records, so views, boards, rules, and approvals all see them.
Can employees use a shared tablet instead of their phones — a cafeteria register, a tool crib?
Yes, if the app holds the kiosk capture grant. Its settings page shows a kiosk link: open it once on the tablet and it becomes the kiosk — tap your badge → pick what applies → Done, and the screen resets for the next person. The tablet’s USB or Bluetooth badge reader “types” the badge number (the same RF IDeas-style readers used for workstation tap-in); a QR-coded badge works with a 2D scanner or the tablet camera, and typing the badge number works too. The kiosk is badge only — an employee number is refused, because a number is printed on paperwork while a badge has to be in hand — and every record it writes names the badge’s owner in the person field and the change history. A badge counts when it matches the Badge ID on the employee’s profile; fill that by tapping the badge on the profile’s Badge ID box, or through the user import. “New link” in the app’s settings unpairs every tablet at once; “Turn the kiosk off” closes the door while the app keeps running. The Cafeteria Meal Log template ships with a kiosk.
Can a badge tap fill in the person on an ordinary record, not just at a kiosk?
Yes. Turn on Identify by badge tap for any person field (the app’s Fields page → edit the field; the composer’s plan carries it as badge: true). The record form then shows a badge box above the people picker — a tap or a camera scan of the badge’s code selects the person, exactly as searching would. Typing a colleague’s badge number in that box works too, so this is a convenience, not a lock; the kiosk is the surface that insists on the badge.
I revoked the public link — is my printed QR code garbage now?
No, and this is deliberate. Revoking public capture kills the link immediately: the form 404s and nothing can be submitted. But the token survives the revocation, so re-granting the capability revives the SAME link — the QR code taped to the gate works again without reprinting. If you want a genuinely new link (say the old one leaked), rotate the token from the app’s settings instead.
Can a mini app require a manager’s sign-off before something proceeds?
Yes, with the request approval grant. The plan names who approves in the same vocabulary as notifications (a role, a person-field’s manager, and so on), and the card spells out what happens on approve and on decline — those follow-up actions are part of what you review. The approval itself is the platform’s real approval flow: the approver gets a proper request with a human-readable summary, and the decision is recorded on the record.
What is the quick-update form, and why can someone with no table access use it?
It is one narrow door: identify a single record by its code, and report a new number for one field. The confirmation names the record and shows the movement — 12 → 9, not just “saved” — and an optional note lands in that row’s change log alongside every other change.
It is safe to expose that widely precisely because it is narrow. It cannot create records, browse or read the table, touch any other field, or be pointed at a different table — every one of those is fixed by the app’s definition, never by the request. So a member who cannot read the app’s records can still open the mini app, use the door, and see an honest notice about what they cannot see, rather than a silently empty grid.
I deleted a view the app created — what happens now?
It stays deleted. What a later change can do depends on where the app came from, and the app’s settings page says which:
- A MangoApps app — one MangoApps ships and keeps updating, not a copy you started from a template — receives those updates, and they only ever add pieces — a later version never brings back something you deliberately removed, and everything else in the app keeps receiving updates as normal.
- An app built in your business — including one you started from a template, which is your own copy — is changed by its admins on its Fields and Rules pages, and with Edit with AI, which can also rename, retype or remove fields, or change whether they’re required — and lists each change on its review card before anything is applied. Nothing else overwrites your renames or added fields, and a field you removed is never brought back.
Renaming anything is always safe (views, rules, and the board reference fields by ID), and archiving an object keeps it resolvable. If you want a removed view back, ask for it in the app’s edit loop (Edit with AI) and approve the change — turning the app off and on again does not re-create it.
What happens to a mini app’s records when I remove it?
They go with it. Remove on the app’s row in Your apps (or on its Edit with AI page, when its setup failed) takes the lists the app created out of service at once, and deletes them and every record in them 7 days later. The confirm says how many records that is. To keep a copy, export them first from the app’s Import & export page, which the app’s administrators and Power Tables administrators can open — the confirm points you there when you can (an app that is turned off is turned on first: its pages don’t open while it is off). Until the 7 days are up, an administrator, or someone with Manager access to a list, can restore it from the Archived tab in Power Tables (Power Tables has to be turned on in Admin → Apps → All Apps) — it comes back as an ordinary table there, not as the app, and the app’s rules on it stay paused until someone resumes them. The message after Remove gives the date, and says you can restore them when you are one of the people who can.
A list the app took over from Power Tables, rather than created, was yours before the app and stays in Power Tables, with the app’s rules on it paused. A mini app that MangoApps ships keeps all its lists, since it can be installed again. Apps removed before this changed kept their lists, and those are still in Power Tables.
If I start from a template, do I get updates to that template later?
No. A template compiles a copy. From the moment you approve it the app is yours, fully editable, with no upgrade tie back to the source — the app’s settings page names the template it came from and says exactly this. The template shelf lists the platform’s templates (fields, views, rules and pages together) and your workspace’s own mini apps that you can open, as a starting point for a copy.
I drafted a mini app but navigated away — did I lose it?
No. Drafts are saved automatically the moment a plan comes back, so drafting does not have to be done in one sitting. Resume or discard them from the Your open drafts card on the compose page. Drafts are per-admin working notes, not a shared surface, and approving a plan settles its draft.
Permissions and access
Why can’t my employees see the custom app I just deployed?
A newly deployed custom app starts unpublished — only you, your administrators, App Builder admins and anyone given a role on the app can open it. To roll it out, press Publish on its row in App Builder → Your apps, or scope it to specific groups with a visibility rule from Admin → Apps → All Apps. Until you publish, employees cannot see it in the app launcher or navigate to it. Mini apps work the same way.
Can employees change settings in a deployed app?
No. Settings inside a deployed custom app are restricted to administrators. Employees can use the app — view records, create records, submit forms — but cannot change its configuration.
Who can change a mini app’s settings?
Someone holding an app-admin grant on that mini app, or a Power Tables admin. Each mini app is a real marketplace app, so a global admin can delegate administration of one specific mini app without handing over App Builder itself. That person can rename the app’s vocabulary, print QR labels, manage its fields, and switch the app off. Who can use the app is a separate question, governed by the mini app’s own audience — not by Power Tables’ audience.
Day-to-day
How do I edit an app that is already deployed?
For a custom app: open the build session from Build history or click Edit with AI on the app in Your apps. Describe the change in the conversation. The builder drafts a plan and pauses for your approval before writing code, the same way it does for a new build. After you review the changes, click Deploy or Update to push them live. Version history keeps the previous state one restore away.
For a mini app: press Edit with AI at the top of any of its pages, which opens on the page you were looking at (or use Edit with AI on its row in Your apps), and describe the change — add, rename or remove. The proposed change renders on its own approval card, and applying it bumps the app’s version. There is no deploy step — the change is live when you approve it.
What does the plan-approval step look like?
For a custom app, the builder drafts a plan listing the record types, screens, files, and capabilities the app will include, then pauses. For a mini app, the approval card lists every table with its fields and who can write to it, each saved view and what it filters to, each automation and what it does, the pages, the board, the quick-update form — and, in their own block, any capability grants the app needs (a public form, outbound email, approvals), each explained in plain words, because approving the card is what grants them. In both cases you can revise, edit, or decline before anything is created, and in both cases the plan is re-validated on the server when you approve it — so what you approved is what you get.
Can I attach screenshots or documents when describing an app?
Yes, in the custom-app builder. You can attach up to 5 images per message, each up to 10 MB. Accepted formats are PNG, JPEG, GIF, and WebP. The mini-app composer is text-only — you describe what you want to track in a sentence or two.
How do I save a build as a reusable template?
After a successful custom-app build, click Save as template in the builder. The template captures the original description and the approved plan specification, and is scoped to your business. You can also create templates manually from the Templates shelf. System templates (the curated gallery) are read-only, but you can import one as an editable draft. Mini-app templates are platform-provided and appear at the top of the same Templates page.
How do people use a mini app from the shop floor?
Two ways, neither of which needs an app install or a scanner. Print QR labels from the app’s Manage menu gives you a sheet of one code per record; scanning a label with a phone camera opens that record’s mobile card with − / + steppers. Or send people straight to the app’s quick-update door, where they type the code and the new number. Both land in the same audit trail as every other change.
When something looks wrong
My app stopped loading — what happened?
This applies to deployed custom apps. When one fails to load, App Builder records the failure. After 3 consecutive failures, the app is quarantined and stops being loaded for users entirely. You will see a health alert in your inbox if alerts are on. From Your apps, you can retry loading the app, or open its build session with Edit with AI and use Fix with AI to repair the code. Rolling back to an earlier deployed version is on the app’s page under Admin → Custom Apps.
The builder hit its iteration limit and paused — what do I do?
Click Continue. Each click doubles the builder’s iteration budget for that session. If Auto-continue long builds is on in Settings (the default), medium and rich builds extend automatically up to 2 extra rounds before pausing. The absolute ceiling is 120 iterations. If that is not enough, the per-session token budget (12 million tokens by default) is the real stop.
I got a “still processing” error when I sent a message
The builder processes one message at a time. If you see this error, a previous turn is still running. Wait for it to finish, or click Cancel to stop the current generation before sending a new message. Short confirmations like “yes” or “build it” sent while a build is already running are absorbed silently so they do not block you.
My mini app says it is still being set up — is it broken?
Almost certainly not. Creating a mini app sets up several objects (a table, its columns, views, rules, and possibly a board), and that work finishes in the background. If a step fails, the platform retries and heals forward rather than starting over — the process is designed so a retry completes the remainder instead of duplicating anything. Give it a minute and reload. If it is still stuck after that, check Your apps.
I see a health alert about an event handler being suspended — what does that mean?
A built custom app’s event handler was suspended after repeated failures when processing events. Events for that handler are being dropped until an administrator unsuspends it. Open the app in the builder or from Admin → Custom Apps to investigate and redeploy a fix.
I can’t find the app on my phone — where did it go?
The custom-app builder itself is desktop-only. On a phone you see a “Building happens on desktop” card and a list of your live built apps that have a mobile view. Not every built app has a mobile page — only apps with record types are servable on mobile. Mini apps work in a phone browser, and QR labels deep-link straight to a record’s mobile card.
I submitted a form through the public link but nobody received it — why?
Public submissions land as records in the app, and a new record tells nobody by itself. For a mini app, ask an admin of the app to add a rule on its Rules page that notifies someone when a record is added — a public submission is an ordinary record, so the rule covers it. For a custom app, the app’s notification rules or SLA configuration may not cover the public-submission flow; ask your administrator to check the app’s notification and overdue settings in the builder.
How many versions of my app are kept?
The builder keeps up to 10 file snapshots per build session. Versions increment as you build and edit. When the cap is reached, the oldest snapshot is dropped. Restoring a version takes a safety snapshot of your current files first, so you do not lose the state you are rolling back from.
Licensing and limits
What are the mini-app limits I should know about?
| What | Limit |
|---|---|
| Tables per mini app | 1–3, linked with reference fields; each table gets its own page |
| Fields per table | 1–20 |
| Field types available | Text, number, counter, choice, date, person, attachment, reference (a link to one record of another table or mini app), lookup (show a field through a reference) |
| First field | Must be text — it is the record’s display name |
| Field name length | 60 characters |
| App name length | 1–60 characters |
| Options on a choice field | 2–20 |
| Saved views per table | 6 |
| Rules per mini app | 10 |
| Rule actions — always available | Update a field, notify a person, create a task, scheduled digest of a saved view |
| Rule actions — capability grants (disclosed on the card) | Send email to an address stored on the record, request approval, public capture form |
| Rule recipients allowed | The record’s filer, the app’s creator, a person-field on the record, that person’s manager, a role, or a team |
| Rule triggers allowed | Record created, field changed, schedule |
| Pages (screens) per mini app | 5, each with up to 5 sections; metric cards per page: 4 |
| Rows shown on a live board | 3–10 |
| Description length when composing | 2,000 characters |
| Description length when editing | 1,000 characters |
| Automatic repair rounds on an invalid draft | Up to 2, and only while each round fixes more than the last |
| Open drafts kept per composer | 10 |
| Records per printable QR label sheet | 120 |
| Public form submissions | Honeypot-protected and rate-capped per form per hour |
| How many mini apps you can run | Bounded by your company’s table limit (each table counts) and at most 25 unpublished at once; live ones are not capped — where mini app packs apply, the pack follows them |
How many mini apps can we have?
Two limits apply. Tables: each mini app creates one to three tables, and every table counts toward your company’s table limit, which it shares with tables built by hand. The default is 5 tables; an administrator can raise it to 25 on Power Tables’ settings page without asking anyone (Power Tables has to be turned on in Admin → Apps → All Apps to open that page). Beyond that requires a MangoApps Console grant, which tops out at 500. If approving a mini app is refused with “This app needs N new tables, and your company’s table limit … has room for M”, that is the number to raise — then approve it again (an AI draft stays in Your open drafts; a plan started from a template is started again from that template). An app composed earlier whose setup stopped on “Table limit (5) reached” recovers the same way, with Retry setup now. Unpublished apps: up to 25 apps can wait unpublished at once; publish or remove one to create another. Live apps are not limited: where mini app packs apply, the pack follows how many are live and never stops a publish.
What are the custom-app build limits I should know about?
| What | Limit |
|---|---|
| Platform APIs available to a built app | 81 |
| Declared capabilities an app can request | 16 |
| Image attachments per message | 5 |
| Image size | 10 MB |
| Builder chat messages | 60 per hour |
| Agent iterations per build turn | 40 |
| Absolute iteration ceiling per session | 120 |
| Wall-clock limit per build turn | 600 seconds |
| Per-session token budget | 12,000,000 tokens |
| Auto-continue rounds (medium and rich builds) | 2 |
| File snapshots retained | 10 |
| Quarantine threshold | 3 consecutive failures |
| Health alert deduplication window | 6 hours |
| API key scopes | builder, query, manage |
| AI-usage log retention | 30 days |
How is App Builder priced?
App Builder is included with every edition, with no per-employee charge and no per-app charge for custom apps. Building and trying apps is always free: an app counts only once it is published to users and working.
Where mini app packs are switched on (our shared cloud), live mini apps are billed as the smallest pack that covers them — 1 app $149 a month; up to 3, $387; up to 10, $1,100; up to 25, $2,250; up to 50, $4,000; up to 100, $7,000; beyond 100, $70 a month for each app. The pack follows your live mini apps on its own: each night it moves to the pack that covers them — up at once, down on the 1st — so there is nothing to buy, and it never stops you publishing. Mini apps already live when packs were switched on stay free, and apps your order form covers are billed on it. Each published mini app counts once, and that includes the app’s records and its boards. Three consequences worth knowing:
- An unpublished mini app is not counted, so build and test as many as you like (up to 25 unpublished at once).
- A mini app that fails to load is free while it is broken.
- Custom apps are not counted.
A mini app’s records belong to the app, not to Power Tables, so they never add to Power Tables’ usage. Builder AI tokens are not billed separately either — they stay on the shared platform AI-credits meter, and the Analytics tab shows the spend. The pack prices above are the list prices; for the rates on your own bill, talk to your MangoApps account team.
Can I build apps from external tools instead of the browser?
Not end to end — building and deploying a custom app happen in the browser. On a private cloud, where coded apps run, App Builder does issue scoped, revocable API keys for a REST API of the builder’s tools (it is not an MCP server). A script or coding assistant can list the tools, call them — read the SDK docs and examples, check code — and run read-only platform queries against your tenant; no key can write an app’s files or deploy it. Keys track last use and can be revoked from the builder at any time. The three available scopes are builder, query, and manage.
What quality checks does a build go through before I can deploy?
Every custom-app build passes three checks before deploy is offered. First, deterministic fit checks validate the schema, paths, and file structure. Second, an AST-level static security scan inspects the generated code for sandbox escapes, forbidden method calls, and tenant-isolation violations. Third, an AI QA reviewer reads the intent (your description plus the approved plan) against the generated files and judges whether each scenario is correctly implemented.
Mini apps do not go through these, because there is no code to scan. Their equivalent safeguard is the palette itself: the plan is validated against a fixed set of allowed field types, operators, actions, and recipients before you see it, and validated again on the server when you approve it.
What can the Ask AI agents do — and what can’t they?
Two agents split this territory, and Ask AI routes to the right one by what you ask.
The App Builder agent covers built custom apps. Four read tools: list the custom apps this business has built, report one app’s health, show a built app’s deploy/enable/disable audit trail, and list recent build sessions. Two write tools — enable an app and disable an app — both confirm first, only work on apps this business actually built, and are global-admin only. It cannot deploy, roll back, or delete an app, and it does not cover mini apps.
The Mini Apps agent answers questions about your mini apps, for anyone who can open them: list the mini apps you can open, describe one’s lists, fields, views and pages, look up and search its records, count them (per list and per saved view), and submit a quick-update form (with confirmation). It enforces each app’s own audience and each list’s access on every question — ask about a list you cannot see and it tells you so, rather than guessing — and an app switched off in its settings is named as switched off.
Settings
What do the App Builder settings control?
Settings is for global administrators only. Besides Enable AI agents (the switch for both App Builder agents in Ask AI) and Who can create mini apps, it carries three toggles:
- Auto-continue long builds (default: on) — when a medium or rich custom-app build hits its iteration limit, the builder extends automatically (up to 2 rounds) instead of pausing for a click.
- Alert on built-app health problems (default: on) — sends a notification when a built app is quarantined, fails to load, has an event handler suspended, or has an auto-update rolled back. Alerts are deduped to at most one per app per event per 6 hours.
- Alert when a marketplace-installed app has an update (default: on) — sends a notification when a newer published version of an app your business installed from the marketplace is available. Apps built here raise no update alert — a built app changes when you redeploy it.
An optional Additional notification recipients field lets you widen alerts to a group from Admin › Groups. Without it, alerts go to the app’s builder and deployer first, then fall back to App Builder’s app-admins.
The three toggles and the group concern custom-app builds. None of them affects mini apps, which have no build job, no iteration limit, and no deploy step.
Where are a mini app’s own settings?
On the mini app itself, on its App settings tab in the app’s sidebar (an admin of the app sees it). That page carries the app’s fields and where they came from, the vocabulary controls (rename “records” to whatever your business calls them, and relabel the tabs), the full list of objects the app manages, and an on/off switch for the whole app. Turning the app off hides it for everyone; the data is untouched and the settings page stays reachable so you can turn it back on.
More help
- App Builder Overview
- Ask AI — the assistant answers App Builder questions from these articles.