Getting Started
This tutorial walks you from nothing to a real pull request that adds the
default profile’s agentic workflows to a repository you own. By the end you
will have registered a repo in your fleet, previewed the change, opened a PR, and
confirmed the repo is aligned.
Everything gh-aw-fleet does to a repository happens through a pull request you
review and merge yourself. Nothing is force-pushed and nothing is committed to
main for you, so this tutorial is safe to follow on a repository you control.
You will need the tools from the Install guide: the gh CLI
authenticated with repo and workflow scopes, the gh aw extension pinned to
v0.79.2, and the gh-aw-fleet binary on your PATH. Set aside about ten
minutes.
Throughout, replace you/your-repo with a repository you own.
Step 1 — Confirm your tools are ready
Section titled “Step 1 — Confirm your tools are ready”Run these three checks:
gh auth statusgh extension list # gh-aw should appear, pinned to v0.79.2gh-aw-fleet listgh-aw-fleet list prints the repos it already tracks and, on stderr, which
config files it loaded — (loaded fleet.json) on a fresh checkout. If all three
commands succeed, you are ready.
Step 2 — Register your first repo
Section titled “Step 2 — Register your first repo”Add your repo to the fleet with the baseline default profile. Run it first
without --apply to see what it would write:
gh-aw-fleet add you/your-repo --profile defaultThis is a dry-run: it prints the planned entry and changes nothing. When the plan looks right, write it for real:
gh-aw-fleet add you/your-repo --profile default --applyThe command asks you to confirm; answer y. It writes the entry to
fleet.local.json — your private, gitignored fleet state — not to the
committed fleet.json. Run gh-aw-fleet list again and your repo now appears in
the table.
Step 3 — Preview the deploy
Section titled “Step 3 — Preview the deploy”Now see exactly what would be installed, still without touching the repo:
gh-aw-fleet deploy you/your-repodeploy is dry-run by default. It resolves the default profile into a concrete
set of workflows, pinned to the fleet’s source refs, and prints the plan. Read
it: these are the workflow files the pull request will add.
Step 4 — Open the pull request
Section titled “Step 4 — Open the pull request”When the plan looks right, apply it:
gh-aw-fleet deploy you/your-repo --applyThis clones the repo to a scratch directory, runs gh aw to add and compile the
workflows, commits them, pushes a branch, and opens a PR. When it finishes it
prints the PR URL. Open it — you will see the added .github/workflows/ files,
ready for review.
If the commit stops on a gpg signing prompt or error, that is expected on signing-required setups. Follow Recover from a gpg signing failure to finish the commit in your own shell, then come back here.
Step 5 — Confirm it worked
Section titled “Step 5 — Confirm it worked”Review and merge the pull request on GitHub. Once it is merged, ask the fleet whether the repo now matches its declared state:
gh-aw-fleet status you/your-repoThe repo reports aligned — the workflows on main match what the default
profile declares. For a fuller picture of the repo’s recent run health and
credit spend, run:
gh-aw-fleet overview you/your-repoWhat you did, and where to go next
Section titled “What you did, and where to go next”You registered a repo, previewed a change, opened a reviewable PR, and confirmed
the repo is aligned — the full gh-aw-fleet loop, end to end.
From here:
- Solve a specific problem — gate CI on fleet drift, resume an interrupted apply, or find your top credit burners.
- Look up a command or flag — the CLI reference lists every command, flag, and exit code.
- Understand the model — Reconcile workflow and Consumption and FinOps explain how and why the tool behaves as it does.