AGI Wings
Workflow automation and internal SaaS,
in production fast.
AGI Wings is a cloud runtime optimized for AI agents. No no-code limits and no complex infrastructure setup. Your AI agent assembles JS, SQL databases, and external service connections and deploys to the cloud in an instant—from scheduled jobs and webhooks to web apps for your team or just yourself.
Copy and paste into your AI agent
Set up AGI Wings (https://agi-labo.com/tools/wings) and create one sample Web App. Follow https://agi-labo.com/en/tools/wings/docs/getting-started. Supported environments: macOS, Linux, and Windows (on Windows, run it in Git Bash or PowerShell; WSL is not yet verified). 1. Install the CLI: `curl -fsSL https://agi-labo.com/wings/install.sh | sh`. In PowerShell on Windows, use `irm https://agi-labo.com/wings/install.ps1 | iex` (if wings is not found, add `~/.agi-tools/bin` on macOS/Linux or `%USERPROFILE%\.agi-tools\bin` for the Windows CLI to PATH). 2. Run `wings login`. When the browser opens, wait while I sign in with my AGI Lab account. 3. Install the Skill with `wings skill install --target all --yes`, check its location with `wings skill status`, and read that SKILL.md. 4. Create a notes Web App that only I can open. Let me write and save notes and view a list of them. Store the notes in a Wings Database. Set a title and description too. 5. Give me the URL to open and three things I can ask Wings to do next.
HOW IT WORKS
You describe it. Your agent writes it. Wings runs it.
Wings runs the code and connects to your services. You choose when it starts and where to save data.
01
You
Describe the job
02
Your agent
Write and deploy the code
03
AGI Wings
Run it in the cloud
export default async function (event) { const res = await fetch('https://api.github.com/issues', { headers: { 'User-Agent': 'github-digest' }, }) const issues = await res.json() const text = issues.map((i) => `- ${i.title}`).join('\n') await fetch('https://slack.com/api/chat.postMessage', { method: 'POST', headers: { 'Content-Type': 'application/json' }, body: JSON.stringify({ channel: event.channel, text }), }) return { count: issues.length }}$ wings connection add github --host api.github.com --auth bearer$ wings connection add slack --preset slack$ wings handler deploy --name github-digest --file handler.js \ --connection github=github --connection slack=slack \ --schedule "every:1d@09:00" --tz Asia/Tokyo \ --args '{"channel":"C0123456789"}'✓ Deployed handler "github-digest"HOW IT FITS
Just the pieces you need.
Choose and combine the pieces your job needs.
Schedule
Run on a schedule
Choose when recurring work should run. Starts may be delayed when the service is busy.
API Trigger
Respond to a request
Receive webhooks and HTTP requests.
Web App
Open in a browser
Share a private app with people you choose.
Handler
Run your code
A JavaScript function does the work.
Connection
Connect your services
Keep API credentials outside your code.
Database
Keep your data
Read and write data with SQL.
Run
Review the result
Check what happened in the Console.
USE CASES
Ask for the work you want.
Install the Wings Skill so your agent can deploy and check the result in the same conversation.
wings skill install --target all --yes
Schedule
A daily summary
Every morning at 9, summarize my GitHub issues and send them to Slack.
API Trigger
A webhook receiver
Receive Stripe payment webhooks, verify their signatures, and notify Discord.
Web App
A shared team tool
Make a private app that lists recent Stripe sales and share it with our finance team.
SECURE BY DEFAULT
Only the permissions it needs.
A Handler can use only the Connections you select when deploying it. Connection credentials never reach the running code.
Read about security and data →- Credentials kept out of code
- Only selected service connections
- Execution history in the Console
GET STARTED
Give your AI agents wings
Start with a free AGI Lab account. Active members get higher limits at no extra charge.
Copy and paste into your AI agent
Set up AGI Wings (https://agi-labo.com/tools/wings) and create one sample Web App. Follow https://agi-labo.com/en/tools/wings/docs/getting-started. Supported environments: macOS, Linux, and Windows (on Windows, run it in Git Bash or PowerShell; WSL is not yet verified). 1. Install the CLI: `curl -fsSL https://agi-labo.com/wings/install.sh | sh`. In PowerShell on Windows, use `irm https://agi-labo.com/wings/install.ps1 | iex` (if wings is not found, add `~/.agi-tools/bin` on macOS/Linux or `%USERPROFILE%\.agi-tools\bin` for the Windows CLI to PATH). 2. Run `wings login`. When the browser opens, wait while I sign in with my AGI Lab account. 3. Install the Skill with `wings skill install --target all --yes`, check its location with `wings skill status`, and read that SKILL.md. 4. Create a notes Web App that only I can open. Let me write and save notes and view a list of them. Store the notes in a Wings Database. Set a title and description too. 5. Give me the URL to open and three things I can ask Wings to do next.
Manual installation
curl -fsSL https://agi-labo.com/wings/install.sh | sh
On Windows, use the command above in Git Bash, or irm https://agi-labo.com/wings/install.ps1 | iex in PowerShell, and add %USERPROFILE%\.agi-tools\bin to PATH. WSL uses the Linux build (not verified). PowerShell 5.1 and 7 require ssh-keygen.exe from Windows OpenSSH.
New Free accounts have lower limits for their first 7 days: 3 Handlers, a 15-minute minimum Schedule interval, and 10 minutes of Run duration per UTC day.