AACWorkflow Docs

10-Minute Quickstart

Sign up, install, create an agent, assign your first task, and link a pull request — everything you need to get started with AACWorkflow in 10 minutes.

This page walks you end-to-end through AACWorkflow Cloud in 10 minutes: sign up → install the CLI → start the daemon → create an agent → create a workspace issue → assign the agent → watch it work → link a pull request. Everything you need to see AI coding tools collaborate with your team.

Prerequisite: You have at least one AI coding tool installed locally (Claude Code, Cursor, Copilot, etc.). The daemon auto-detects them on startup.

1. Create a Cloud account (1 min)

Sign up at aacworkflow.com with email or Google OAuth.

After sign-up, you're automatically placed in a default workspace. You can rename it later or create new workspaces.

2. Install the CLI (1 min)

macOS / Linux (Homebrew):

brew install aacworkflow-ai/tap/aacworkflow

Windows (PowerShell):

irm https://raw.githubusercontent.com/aacworkflow-ai/aacworkflow/main/scripts/install.ps1 | iex

Verify:

aacworkflow version

3. Log in and start the daemon (2 min)

One command handles both:

aacworkflow setup

This will:

  1. Open your browser for login (same email verification / OAuth as the web)
  2. Store your API token in ~/.aacworkflow/config.json
  3. Start the daemon — it begins polling for tasks

Verify it's running:

aacworkflow daemon status

You should see online.

Using the desktop app? The desktop app starts the daemon automatically on launch — no CLI setup needed.

4. Verify runtime is online (1 min)

In the web UI:

  1. Go Settings → Runtimes
  2. You should see your daemon listed with one or more active runtimes (one per AI tool detected)

If it shows offline, see Troubleshooting.

5. Create your first agent (1 min)

In the web UI, go Settings → Agents → New Agent:

  • Name — give your agent a name (e.g., code-reviewer)
  • Provider — pick an AI tool you have installed (dropdown only shows detected tools)
  • Model (optional) — select a model inside that tool
  • Instructions (optional) — a custom system prompt for this agent

Click Create. Your agent now appears in the member list and can be assigned work.

6. Create an issue (1 min)

In the web UI, click New Issue (or use the CLI):

aacworkflow issue create --title "Add ASCII architecture diagram to README"

You'll see the issue appear on the board with status queued.

7. Assign the agent (1 min)

In the web UI, open the issue and click the assignee field. Select your agent.

Or use the CLI:

aacworkflow issue assign AAC-1 --to code-reviewer

What happens next (watch in real-time):

  1. Within 3 seconds: status → dispatched (daemon picked it up)
  2. Next 5–30 seconds: status → running (agent is working)
  3. Agent: reads your code, runs commands, edits files locally
  4. When done: status → completed (or failed if retry policy kicks in)

The web UI updates live via WebSocket — no refresh needed.

If your agent created a pull request or branch, click Link PR in the issue's top-right corner and paste the GitHub/GitLab URL. AACWorkflow will:

  • Display the PR status in the issue
  • Sync PR comments to the issue comment thread
  • Mark the issue resolved when the PR merges

Done! You've just completed a full AI-driven task loop — from issue creation through agent execution to PR linking. Now explore:

  • Squads — group agents by skill (e.g., frontend, backend, QA)
  • Autopilots — set standing rules (e.g., "route all Terraform diffs to the DevOps agent")
  • Chat — chat with agents inside issues
  • Skills — attach knowledge packs to agents

Next steps

  • Cloud quickstart — more detail on daemon setup and troubleshooting
  • Agents — understand how agents differ from humans
  • Tasks — task lifecycle, retries, and failure modes
  • Daemon and runtimes — how the daemon operates
  • AI coding tools — capability matrix across Claude Code, Cursor, Copilot, and 9 others

On this page

1. Create a Cloud account (1 min)2. Install the CLI (1 min)3. Log in and start the daemon (2 min)4. Verify runtime is online (1 min)5. Create your first agent (1 min)6. Create an issue (1 min)7. Assign the agent (1 min)8. Link the pull request (2 min)Next steps