Quickstart

This page takes you from “nothing installed” to “an agent has closed an issue” without leaving the docs. Plan on five minutes.

What you’ll need

• Operating system: Windows 10 or newer (64-bit). The macOS build is in progress — see agentslan.com/download.
• At least one supported coding-agent CLI installed on your machine. Agentslan is a control plane; it doesn’t ship its own model. Pick one of:
  • Claude Code — Anthropic’s official CLI. The smoothest first install.
  • Codex CLI — OpenAI’s official CLI.
  • Cursor — Cursor’s headless Agent CLI.
  • Gemini CLI — Google’s CLI.
  • OpenCode or Pi — multi-provider routing CLIs.
  • OpenClaw — separate self-hosted gateway.
• An Agentslan license key. Buy a plan at agentslan.com/pricing (Solo includes a 14-day free trial). The licensing server emails the key on checkout.

1. Download and install

Grab the Windows installer from agentslan.com/download. The installer is about 61 MB and finishes in under a minute.

Handling the SmartScreen prompt

The current beta build is not yet code-signed, so Windows will show a blue “Windows protected your PC” dialog the first time you run the installer.

1. Click More info (small link at the top of the dialog).
2. Click Run anyway.
3. The installer takes about 30 seconds. Launch Agentslan from the Start menu when it finishes.

The installer is the same file the Agentslan team builds and uploads. The warning only appears because the code-signing certificate has not been provisioned yet.

2. Activate your license

On first launch the desktop app prompts for your license key. Paste the key from the email Stripe + the licensing server sent at checkout and confirm.

The app then verifies the key offline using a built-in Ed25519 public key — no network round-trip is needed for ongoing use. If the key has expired or is malformed the app explains exactly what’s wrong; treat any other error as worth a support@agentslan.com email.

3. Connect your first agent

We’ll use Claude Code as the example because the install is the shortest; swap any other adapter from the list above if you prefer.

1. Install Claude Code on the host (one-off):

   npm install -g @anthropic-ai/claude-code

2. Sign in to Claude Code in your terminal so its credentials are cached on disk:

   claude

   Complete the OAuth flow in your browser. Type /exit when done.

3. Add the agent in Agentslan:

   • Go to Agents → Add agent.
   • Pick Claude Code (local) as the adapter.
   • Choose a model (start with claude-sonnet-4-6 — fast, cheap, capable).
   • Save.

For the full configuration reference see the Claude Code adapter page.

4. Dispatch your first issue

1. In the desktop app, create a project pointing at the local repository you want the agent to work in.
2. Create a new issue with a small, well-scoped task — for example “Add a LICENSE file with the MIT license, attributing my company”.
3. Assign the issue to the agent you just created.
4. Click Dispatch.

Agentslan spawns Claude Code in a sandboxed Git worktree of your project, streams the run transcript into the issue, and updates issue status as the agent works. When the agent finishes, the diff is attached to the issue and you can approve or push it from the dashboard.

Where to go next

• Connect more agents — the same install-then-add flow works for every supported runtime.
• Routines — schedule recurring dispatches (nightly flake-sweeps, weekly dependency upgrades, etc.).
• Shared memory — how agents persist context across runs so they stop relearning the same things.
• Plan, seats, and licensing — managing subscriptions, adding teammates, and what happens if a subscription lapses.
• Troubleshooting — first stop when something gets stuck.