Quickstart

Boardwalk runs agent workflows: TypeScript programs that call models, run on schedules, and keep a full record of every run. This page takes you from nothing to a deployed workflow in about five minutes.

Install the CLI

npm install -g @boardwalk-labs/cli

Verify the install with boardwalk --version. The CLI is open source (MIT); no account is needed until you deploy.

Create a workflow

boardwalk init scaffolds a project from a template. Start from morning-digest, a cron workflow that summarizes your open GitHub issues every weekday morning:

boardwalk init my-digest --template morning-digest
cd my-digest
npm install

A workflow is one TypeScript file. The meta literal declares the contract (schedule, secrets, budget); the rest of the file is the script that runs. See Writing workflows for the full model.

Run it locally

Put your secret in .env, then execute the file on the spot. The run log streams to your terminal; no deploy, no account:

cp .env.example .env   # fill in: GITHUB_TOKEN
boardwalk dev .

Deploy it

boardwalk login opens the browser to authenticate. Then boardwalk run deploys the workflow and triggers a run right away, streaming the status back; from then on the cron trigger fires it on schedule.

boardwalk login
boardwalk run . --org your-org

Your secrets move to the vault in the dashboard (Settings → Secrets), and every run gets a live tail and a permanent record.

Where to next

• Concepts: the five nouns the docs use.
• Writing workflows: the program model and the SDK surface.
• Inference: how agent() picks a model: managed, any model, or your own key.
• Triggers and Secrets: wiring up schedules and credentials.
• Self-hosting: run the whole thing on your own hardware with one docker run.
• CLI, Manifest, and SDK reference: every command, field, and export.