First step · CLI

Start with a local project.

Install the CLI and create project context. Running a task through the service is a separate next step.

You will need

Node.js 22.12+, your repository and a configured agent CLI for task execution. No sudo is required.

curl -fsSL https://0dai.dev/install.sh | sh first task walkthrough

7 terminal CLIs + Cursor MCP bridge · local initialization works offline · cancel keeps what is in your repo

The npm package is frozen and no longer updated — install.sh is the supported path.

Install the CLI

0:00 – 0:30

One shell command installs the CLI. It needs Node.js 22.12+, asks for no sudo, and drops the launcher in ~/.local/bin/0dai. Every command below then runs from any repo on your machine.

$ curl -fsSL https://0dai.dev/install.sh | sh
$ 0dai --version

0dai CLI is on your PATH — the installer prints the PATH line if ~/.local/bin/0dai is not on it yet. Source stays on your machine.

Init the project layer (no account)

0:30 – 1:30

Run init from the root of any repo. 0dai detects your stack, your installed agent CLIs, and writes the ai/ layer plus CLAUDE.md and AGENTS.md. The remaining per-agent configs need the service-backed setup. No signup is needed for this local scaffold.

$ cd your-project
$ 0dai init --local

ai/ plus CLAUDE.md and AGENTS.md are in place. Review the remaining-config notices.

Inspect the layer with status and doctor

1:30 – 2:00

0dai status is a read-only check of your local ai/ layer, privacy mode, and daily quota. 0dai doctor inspects ai/ manifests and installed agent CLIs, reporting any missing files or remaining per-agent configs until you run 0dai sync.

$ 0dai status
$ 0dai doctor

Review the read-only layer summary and health-check warnings.

After local initialization

You can inspect project files without an account. A run with --now needs the 0dai planning service, your installed agent CLI and its configured provider.

First task walkthrough

Service access and task execution

Run a task through the planning service

2:00 – 4:00

0dai run with --now contacts the 0dai service to plan tasks, then executes them with your installed agent CLI. Service access and the agent provider must be available. Review the resulting receipt and any warnings.

$ 0dai run "add a test for the auth module" --now

After execution, review the run receipt. If it could not be saved, the CLI prints a warning.

Activate the free plan (optional)

4:00 – 5:00

Sign in to activate the free plan when you need service-backed tools. Local initialization needs no account. Check the plan and feature status before relying on cloud or shared-memory capabilities.

$ 0dai auth login
$ 0dai activate free

Check activation status and the tools available to your plan. Your local project layer stays in the repo.

View illustrative install output

~/your-project $ example install
$ curl -fsSL https://0dai.dev/install.sh | sh
0dai downloading the CLI bundle from https://0dai.dev/downloads
0dai checksum ok
0dai v4.10.2 installed: ~/.local/bin/0dai
$ export PATH="$HOME/.local/bin:$PATH"
$ 0dai init --local
8 files generated (ai/ layer + CLAUDE.md, AGENTS.md).
0dai initialized — 5 agent config(s) not written (run 0dai auth login && 0dai sync)
$ 0dai status
[0dai] v4.10.2 | stack: unknown · privacy: local-only
$ 0dai doctor
ai/ layer: ok ai/VERSION · ok ai/manifest/project.yaml · missing ai/manifest/commands.yaml
$ 0dai run "add a test for the auth module" --now
receipt: <run receipt path>

// illustrative output recorded on a clean local project · review actual command results and warnings

Choose what to verify next

After initialization, inspect your project layer. After a task, review its actual outcome and receipt. These paths explain what to verify before you ship:

Read a real receiptSee what a scored run looks like in production First task walkthroughThe longer doc — what to watch for, what to skip PricingCompare platform access and feature status, separately from model and host costs.