Find the guide for a task #
On this page
Choose a task below to open its guide. If something fails, start with the matching troubleshooting page: Setup, Deploys, Secrets, and Status and logs.
Each guide explains the procedure and how to check its result. Your AI tool should read the guide before making changes.
The Quick reference lists the calls each common task takes, and where each key limit is stated.
Tasks and their guides #
The most common tasks #
| Task | Guide |
|---|---|
| Connect your AI coding tool and sign in | Connect your AI assistant |
| Build a new app or bring existing code, test in development, and publish | Build, test and publish an app |
| Deploy an application to production | Deploy from your computer |
| Build a service that keeps its data in a database | Build a database-backed service |
| Let your app's users sign in | Add sign-in to your app |
| Publish a site of HTML, CSS and images as a serverless application | Publish a static site |
| Serve a web client beside the backend, with its fonts, images, and third-party libraries among its files | Serve a web client |
| Build and deploy an app from a chat | Deploy from a chat |
| Choose what to do from a chat without a terminal | Build from a chat AI (no terminal) |
| Read the scaffold templates and their complete source files | Example applications |
| Store an API key or another secret | Store a secret |
| Call an AI model with no provider key of your own | Call an AI model under the allowance |
| Build a public AI summarizer that keeps notes and summaries in a database | Build a public AI summarizer with saved history |
| Run AI calls on your own Anthropic, OpenAI, Google, or xAI key | Use your own AI keys |
| Build a mobile app's backend | Build a mobile app's backend |
Two explanation pages answer the questions most deploys raise: A deployed app's settings and The deploy health check.
Set up, deploy, and run #
| Task | Guide |
|---|---|
| Check what your computer needs | What your computer needs |
| Bring an application built elsewhere | Bring an existing application |
| Build, test, and deploy with nobody at the keyboard to sign in | Deploy with nobody at the keyboard |
| Start a new service from a template | Starting a new service, in The turnzero command |
| Look up the options of a turnzero command line, or the status it ended with | turnzero command reference |
| Look up what a turnzero command line reads or prints | turnzero command input and output |
| Write or change the manifest (the JSON file that describes your application: its services, health path, and other settings), and fix what a refusal reports | Author the manifest |
| Run the application on your machine | Run your app on your computer first, for the environment file and settings (environment variables). Then Test your application locally for the tests. |
| Write a local run's development credentials into the environment file | Running on your machine, in The turnzero command |
| Test each version in development, then promote it to production | Test in development, then promote |
| Roll back, halt, rename, or delete an environment or an application | Manage versions and environments |
| Keep a WebSocket connection open for live updates | Use WebSockets |
| Put the application on a domain of your own | Attach your own domain |
| Run work on a schedule | Add a scheduled job first, for the steps. Then Run scheduled work, in Author the manifest, for the declaration and the runs. Then the Schedule package page, for what the handler must do. |
Data, files, and outside services #
| Task | Guide |
|---|---|
| Add a database to an application you already have | Add a database |
| Build a public form with saved database entries and routes to read, change and delete them | Build a form backed by a database |
| Store each user's files, and let a browser or a native app upload or download one with a transfer grant (a short-lived permission for one named file) | Store files |
| Give the backend a secret's value without pasting it into chat | Keep a secret value out of chat |
| Call an external API with a key you stored | Call an external API with an API key |
| Generate a picture with no provider key of your own | Generate an image with AI |
| Have your AI tool draw one picture now and save it, with no code | Generate an image with your AI assistant |
| Receive a provider's signed webhook | Receive a signed webhook |
| Save signed webhook deliveries once in a database, check development, and promote | Build a webhook receiver with a database |
| Email a person from an application's backend | Send email from your application |
| Read logs and counters, and erase old records | Read logs and counters |
Sign-in and users #
| Task | Guide |
|---|---|
| Let your app's users send feedback, and read their reports with your AI tool | Let your app's users send feedback |
| Tell which user sent a request | Tell which user sent a request, in Add sign-in to your app |
| Keep an admin route, such as deleting a message, for yourself alone | Let only you reach an admin route, in Add sign-in to your app |
| Choose who may sign up, or show a private beta's notice page | Make sign-up invitation-only |
| Add work-account or Apple sign-in | Add work-account or Apple sign-in |
| Invite, list, suspend, or delete end users (the people who sign in to your app), and change the settings of their realm (the list of users for one environment of your app) | Manage end users |
| Test sign-in with no person signing in | Test sign-in with test end users |
| Build an app whose signed-in users keep private notes in a database | Build a notes app with sign-in and a database |
| Build an app whose signed-in users have an AI conversation | Build an app with sign-in and AI |
Mobile and desktop apps #
| Task | Guide |
|---|---|
| Sign in the users of a mobile or desktop app | Sign in from a native app |
| Turn away old versions of a mobile app | Keep installed clients compatible |
| Send push notifications | Send push notifications |
Account, size, tokens, and the library #
| Task | Guide |
|---|---|
| Mint a token for a script, a pipeline, or a contractor's tool | Mint a token |
| Call a management action from a script or shell with no MCP tools | Mint a token |
| Sign a script that has no MCP client in to one of Turn Zero's own products | Sign in from a script |
| Change the size, or read usage | Size and usage |
| Compare where each size runs production, how it starts after idle, and its CPU and memory | How the sizes differ technically |
| See which account your AI tool acts for | Your account |
| Link a second sign-in, or register a passkey | Identities and passkeys |
| Export your account, or delete it | Export or delete your account |
| Hand an app's code and exported data to another account | Hand an app to another account |
| Add or update a reusable feature package from the library | Use the library |
| Report a problem with the platform | Feedback |
| Keep an issue list for a repository or a test program | Keep your own issue list |
Testing and networks #
| Task | Guide |
|---|---|
| Run the tests offline on the packages' test doubles, which stand in for the packages so the tests need no account or connection | Test your application locally |
| Find out why the app loads on one network and not another | When a network blocks Turn Zero |
Files or a database for a list of records #
You can keep a list of records as a JSON file in a storage area or as rows in the database:
- A file suits a small list the application reads and writes whole, such as its settings or a scheduled handler's state. The application should write each change with a conditional put, which saves only if the file is unchanged since it was read. This ensures one writer never overwrites another's change.
- The database suits a list that grows, that the application searches, sorts, or pages through, or that several requests change at once. A guestbook or link shortener fits when its list grows and several requests can write at once. Build a form backed by a database gives a worked example.
- Files also suit content sent unchanged, such as images, uploads, and exports.
Store files covers the conditional put, and Add a database covers the database.
The skills #
A skill is a set of steps offered as a Model Context Protocol (MCP) prompt. Some AI tools show it as a slash command. If your AI tool shows no prompts, it reads a platform skill with read_context, id skill:<name>. Each platform skill links to a guide with examples.
For another product your connection offers, first read its skills:<product> entry to find its skills. Then read one with read_context, id skill:<prefixed-name>. That read sends no arguments; a client that shows prompts can run one with arguments itself. How an AI reads these docs explains the entries and refusals.
The library provides reusable feature packages for capabilities such as sign-in, storage, and AI calls, with integration instructions and, where provided, code to include in your app.
| Skill | Purpose and guide |
|---|---|
skill:deploy |
Deploy from your computer |
skill:store-key |
Store a secret |
skill:add-service |
Add a database |
skill:add-sign-in |
Add sign-in to your app |
skill:manifest-author |
Author the manifest |
skill:use-ai-allowance |
Call an AI model under the allowance |
skill:mobile-backend |
Build a mobile app's backend |
skill:schedule |
Add a scheduled job |
skill:move-existing-app |
Bring an existing application |
skill:develop-then-promote |
Set up a development environment, deploy changes there, then promote them to production. Guide |
skill:release |
Promote a version already in development, check production, and roll back if you request it. With no development environment, check the version deployed directly to production. Guide |
skill:diagnose |
Read logs and counters |
skill:cost-estimate |
Stay inside the Free Allowance |
skill:cost-review |
Size and usage |
skill:custom-domain |
Attach your own domain |
skill:add-library-entry |
Choose a feature package from the library and add it to your project. Guide |
skill:update-library |
Check for newer versions of the packages in your project, ask which to update, and apply their migration steps. Guide |
skill:delete-account |
Export or delete your account |
A nineteenth, platform-admin, is offered only to platform staff. MCP resources and prompts lists the actions used by the customer skills in the table above.
A short skill is one page. A long skill has a first page with its steps and rules, and a page of the reasons behind the rules: skill:add-service, skill:deploy, skill:mobile-backend, skill:release, and skill:schedule are long. The skill:deploy skill also has a page for each case, such as an application with a database. The skill:add-service skill has one for running the application on your own machine. The skill:release skill has skill:release:verify for checking a release.
Some skill rules include a separate line explaining when they apply. Each first page lists the skill's other pages by id.