Quickstart

This walks through the primary flow: sign up on the dashboard, connect GitHub, install the CLI, log in, start the runner, and label an issue to watch it run.

1. Sign up on the dashboard

Go to app.useagentrail.com and create an account (or sign in if you already have one).

2. Connect GitHub

On the Connectors page, authorise GitHub and select the repos you want AgentRail to watch. Set the trigger label to ready-for-agent and choose a poll interval. AgentRail will enqueue any issue you label ready-for-agent.

3. Install the CLI

npm install -g @useagentrail/cli

This installs the agentrail command globally. Verify your install:

agentrail doctor

4. Log in

agentrail login

login runs an OAuth device flow that links this machine to your workspace and saves credentials to ~/.agentrail/credentials.json. Login is required — the only exempt commands are login, logout, whoami, and --help.

Check who you're signed in as at any time:

agentrail whoami

5. Start the runner

agentrail runner

The runner is the local worker that polls your workspace for queued issues, executes them on your machine with your own LLM keys, opens PRs, and reports outcomes back to the dashboard. Keep it running in a terminal (or as a background service) while you work.

agentrail runner --concurrency 2   # run up to 2 issues in parallel

6. Label an issue — get a reviewed PR

On any connected GitHub repo, add the ready-for-agent label to an open issue. The dashboard enqueues it, the runner picks it up, runs a bounded plan → execute → verify loop, and opens a PR. Review the PR and merge.

Next steps

• Understand what's happening under the hood in Core Concepts.
• See the full command surface in the CLI Reference.
• Explore the Dashboard & hosted runner for queue management, cost tracking, and team settings.