Quick reference #
This page gives the calls each common task takes, in the order your AI coding tool makes them. Each row links the guide that explains the steps, their options, and their refusals. Each key limit below links the section that states its figure or, for a figure your plan sets, says where to read it.
Common tasks #
| Task | Calls, in order | Guide |
|---|---|---|
| Connect Claude Code | claude mcp add --transport http turnzero-cloud https://turnzero.ai/mcp, then sign in in the browser, then read_account |
Connect Claude Code or Codex |
| Set this computer up for deploys | npm install -g turnzero, then turnzero login |
Authorize this computer for deploys |
| Create and deploy an application | create_application, submit_manifest, deploy, then run the line it returns |
Deploy an application |
| Deploy from a chat that cannot reach the platform | deploy_files for each file, deploy_files naming those files, then read_status |
Deploying from a chat |
| Turn on a development environment | create_environment for development, then deploy and run the line it returns |
Turn the development environment on |
| Promote a tested version to production | read_status for development, then promote with a wait |
Promote to production |
| Roll back production | list_versions, then roll_back with a wait |
Roll back |
| Store an outside API's key | store_secret, run the line it returns, then declare_upstream |
Store an API key |
| Replace a secret your code reads | rotate_secret, run the line it returns, restart_application, then read_status |
Replace a value |
| Add a database | submit_manifest with the database service, then deploy and run the line it returns |
Add a database |
| Add a scheduled job | submit_manifest with a schedule, deploy, run_schedule, then read_schedules |
Add a scheduled job |
| Call an AI model | list_library, read_library_entry for the AI Allowance entry, take it, generateText from a route, then submit_manifest and deploy |
Call an AI model under the allowance |
| Generate an image | list_library, read_library_entry for the AI Allowance entry, take it, a generateContent request from a route, then submit_manifest and deploy |
Generate an image under the allowance |
| Add sign-in | submit_manifest with the accounts service, a link to /__account/signin, then deploy |
Add sign-in to your app |
| Attach your own domain | submit_manifest with the domain, write the CNAME record, read_status for the TXT record and write it, then read_status until the domain serves |
Attach your own domain |
| Find recent errors | read_status, then read_logs at the error level |
Find the last hour's errors |
| Read usage and change the plan | read_usage, read_plan_quotas, set_plan, then restart_application where the answer names it |
Plan and usage |
Key limits #
Each line says what a limit bounds and links the section that states its figure or says where to read it. Plan figures and prices can change without a release, so read them there or with read_plan_quotas.
- How long one request to your app may run: The execution window.
- How long one scheduled run may run: Schedule.
- The largest request body your app receives: The request size limit.
- The largest file one storage write takes: Limits and metering, in Store files.
- How many requests one address or one signed-in user may send your app a minute: The per-address and per-user limits.
- How many calls a minute the platform takes from one address, and from one credential on the storage routes: The platform request limit.
- How much outbound traffic your app may send: Limits per plan.
- How many database connections each process may hold: The connection limit.
- The largest deploy zip, and how many files it may hold: Names and limits.
- How large a deploy through
deploy_filesmay be: Deploying from a chat. - How many web client files a version may serve, and how large: Keep names and sizes within the limits.
- How long a WebSocket connection may stay open: Limits and metering, in Use WebSockets.
- How many log bytes each environment keeps: Stay under the retained-bytes limit.
- Each plan's quantities and price: What a plan includes.
Related #
- Find the guide for a task links the one guide for every task, the ones above among them.
- Sign-in, sessions, and tokens compares every credential in one table: who uses it, what it may do, how long it lasts, and what ends it.
- How the turnzero command finds its credential says which credential each of the command's lines runs under, the provision line among them.
- Refusals gives each refusal's cause and remedy.
- Glossary defines the terms these guides use.