BY · TORCHSTELLAR

From product idea to
launch-ready project.
Start with one command.

Give AI a real product foundation: website, backend, docs, environment config, and deployment flow, ready from day one.

$curl -fsSL https://1cli.dev/install.sh | bash

Start here

Three ways to start.

Run one create for the fastest path, open templates to choose a stack first, or install the official AI rules and hand the task to an agent.

Option one

Run one create

Use this when you want a project root first. Run one create, follow the prompts for directory and env mode, then use one add to keep adding web apps, APIs, or docs.

Copyable command

$ one create

No flags to memorize first

Prompts ask for directory and env mode

Use one add later for more projects

Explore one create

Option two

Open the templates page

Templates are already grouped by websites, APIs, docs, mobile apps, and desktop apps. Pick a ready example or build a custom mix, then copy the generated One CLI command.

Template page

1cli.dev/en/templates/

Browse by product type

Use ready examples or custom mixes

Run the command yourself or hand it to AI

Open templates

Option three

Hand it to AI

Run one skills install first to install One CLI rules into local AI coding tools such as Codex, Claude Code, and Cursor.

one-line agent prompt

Please use One CLI to create a mobile project named media-stack and install dependencies.

Describe what you want to build

AI creates or migrates projects using One CLI rules

Dependencies are handled by project type

Explore AI guide

Common commands

Complex project setup, split into simple steps.

Humans can run commands directly and follow prompts; scripts and AI can add template names, project names, --yes, and -o json.

one create one add one configure one templates one serve one skills

ONE CREATE

Start a project
with one command.

one create can run directly as a prompt, or run with a directory and the --preset generated by the templates page to scaffold selected web apps, APIs, docs, and libraries.

PROMPTED CREATERun one create by itself to choose the target directory and environment mode.

PRESET CREATEThe templates page turns your selection into one create my-app --preset ... .

AUTOMATIONCI and AI use --yes and -o json for predictable output and errors.

Explore one create Open templatesone-cli/create

$one create

>opens the create wizard

·asks for target directory and env mode

·writes project metadata, automation, and local dev entry

ONE ADD

Keep building
after the project exists.

one add is for an existing One project. You can run one add directly to use the picker, or pass the template name, project name, and deploy option in scripts.

PROMPTED ADDRun one add to choose the template, project name, and optional deploy backend.

AUTOMATIONCI and AI use one add nextjs-app --name web --yes.

SYNC DEFAULTSTemplates can add the expected run, deploy, and AI guide files.

Explore one addone-cli/add

$one add

>checks that cwd belongs to a One project

·opens template and project-name prompts

·writes the new directory and records it in the project list

ONE CONFIGURE

Keep secrets
in the right place.

one configure saves the env, deploy, and container accounts this machine can use. Run it directly for the wizard; scripts use full paths such as env/infisical, deploy/*, or container/docker.

LOCAL FILESWrites under ~/.config/one; secret files are readable only by you.

REMOTE ENVRemote env means Infisical Cloud or your own Infisical. You do not run a separate One CLI service.

REUSED LATERone env, one run, and one deploy read the default profile automatically.

Explore CLI referenceone-cli/configure

$one configure

>open env / deploy / container profile prompts

·save local profile and credential files

·feed later one env, one run, and one deploy commands

ONE TEMPLATES

Pick a starting point
without guessing.

one templates prints the bundled template list in a human-readable terminal format, so people can see the available backend, frontend, docs, mobile, desktop, and library templates first.

TEN TODAYBackend, frontend, mobile, desktop, docs, and TS / Go libraries are included.

HUMAN READABLEBare output shows template names, categories, and what each template is for.

TWO USESTemplate names go to one add; template-page combinations go to one create.

Explore templates Open template pageone-cli/templates

$one templates

>lists templates you can use directly

·shows template ID, category, and purpose

·copy a template name for one add

ONE SERVE

Fill sensitive config
in your browser.

one serve starts a loopback-only local config page for the same profiles managed by one configure. It is meant for human secret entry, so cleartext credentials do not need to pass through an agent chat.

LOCAL ONLYUses a random local port by default and only accepts your own browser, not LAN access.

ONE TOKENThe printed URL includes a session token; old URLs stop working after the process exits.

REMOTE HOSTSRun one serve on the remote machine, then use ssh -L port forwarding from your laptop.

Explore one serveone-cli/serve

$one serve

>opens the local config page in your browser

·URL includes a one-time access token

·Ctrl-C stops the process and expires old links

ONE SKILLS

Help AI understand
how this project works.

one skills install installs or refreshes One CLI's bundled workflows for local AI coding tools. Agents read the project facts first, then create, add, migrate, or bootstrap dependencies by context. Bare install lets you choose tools; --yes installs to every detected tool.

CREATE OR MIGRATECreate a new project from your description, or bring an existing one under One CLI management.

CHOOSABLEBare install opens a picker; --yes installs to every detected tool.

BOOTSTRAP DEPSHandle dependencies by project type: JS/TS uses the package manager; Go uses each subproject.

Explore AI guideone-cli/skills

$one skills install

>detects local AI coding tools

·lets you choose where to install

·syncs command, template, and dependency rules

AGENT-READY WORKSPACE

Let agents read facts before touching the project.

One CLI is not another chat surface. It organizes skills, manifests, AI guides, and structured errors into one engineering contract so Codex, Claude Code, and Cursor can confirm project facts before choosing commands, editing code, or recovering from failures.

Skills explain the workflow

`one skills install` installs One CLI workflows into local AI coding tools. Agents can create workspaces, add templates, migrate projects, or bootstrap dependencies from instructions instead of guessing from a prompt.

one skills install

Manifest records project facts

`one.manifest.json` records projects, templates, toolchains, and environment boundaries. Agents read those facts before choosing a subproject or command.

one.manifest.json

AI guides follow templates

After a successful `one add`, One CLI creates or updates managed `AGENTS.md` / `CLAUDE.md` blocks with template-specific engineering rules. Existing custom content is not overwritten blindly.

AGENTS.md / CLAUDE.md

Errors include next steps

Commands support JSON output, and errors include `code`, `context`, and `remediation[]`. Agents can choose a recovery action without parsing free text.

error.remediation[]

Next steps for every command

When a command fails, you do not have to guess.

From creating and adding projects to running and shipping them, One CLI explains where you are stuck and what to do next. Beginners can follow the message, and AI tools can continue from the same result.

Create projects

Wrong folder or existing target: it tells you how to continue

Add projects

Template, name, or location issues come with a fix command

Run and ship

Missing config or tools are called out before you continue

Connect AI

The same result can be emitted as JSON for tools

>one add --name web --yes

No One CLI workspace found. Run this from a workspace root.
  - This folder is missing one.manifest.json.
    Create a workspace first, or cd into an existing one:
    one create <dir>

This is one example. Create, add, run, and ship commands use the same pattern when something needs attention.

Stop hand-assembling the project foundation.

Install one Go binary, choose templates, and hand clear project context to humans and AI agents.

Start building GitHub

From product idea tolaunch-ready project.Start with one command.