Skip to main content

Prerequisites

  • A Nen API key — get one from the Dashboard
  • An LLM API key (e.g., Anthropic)
  • Python 3.9+

1. Install the Nen CLI

Download the latest release for your platform from releases.getnen.ai, extract it, and add the nen binary to your PATH:
curl -sL https://releases.getnen.ai/nen/latest/nen_darwin_arm64.tar.gz | tar xz
sudo mv nen /usr/local/bin/
Verify the installation:
nen --help

2. Set Your API Key

export NEN_API_KEY="sk_nen_..."
Add export NEN_API_KEY="..." to your shell profile (~/.zshrc, ~/.bashrc) so it’s always available.

3. Create a Desktop

nen desktop create
The command outputs the desktop details including the desktop ID (e.g. dsk_abc123def456):
Desktop created successfully!

  Desktop ID:   dsk_abc123def456

  Status:       running
  Created:      2026-03-31T12:14:20-07:00
Save the desktop ID — you’ll use it in the next step.
Use nen desktop list to see all your desktops and nen desktop status <id> to check a specific desktop’s state.

4. Scaffold an Agent Project

Use the CLI to scaffold a ready-to-run agent project:
nen example create anthropic-computer-use
cd nen-anthropic-cua
Run nen example list to see all available templates.

5. Configure and Run

Install dependencies and add your API keys:
uv sync
cp .env.example .env
Edit .env and fill in your keys:
.env
NEN_API_KEY=sk_nen_...
ANTHROPIC_API_KEY=sk-ant-...   # or OPENAI_API_KEY for the OpenAI template
Run the agent — it will create a desktop, execute the task, and clean up automatically:
uv run python agent.py --task "Open Firefox and navigate to google.com"
Use --desktop-id to target an existing desktop, --keep to skip auto-cleanup, or --max-steps to control the loop length. Run uv run python agent.py --help for all options.

6. Clean Up

The scaffolded agent deletes its desktop automatically when the task finishes. If you used --keep or created a desktop manually, delete it with:
nen desktop delete <desktop_id> --force

Next Steps

CLI Reference

Full CLI command reference

Supported Tools

All tools and supported models (Claude, GPT, and more)

Full Examples

Complete, production-ready agent loop examples

Cloud Desktops

Cloud desktop provisioning, lifecycle, and the tools API