Skip to Content
DocsGuidesAutomation & Scripting

Overview

Since AIDEN is an Electron application, it supports the Chrome DevTools Protocol (CDP) for automation. Using tools like agent-browser , you can script interactions with AIDEN — automating project creation, story management, testing, and more.

This is useful for:

  • CI/CD pipelines — Run automated tests against your AIDEN setup
  • Screenshot generation — Keep documentation screenshots up-to-date
  • Batch operations — Script repetitive workflows
  • Integration testing — Validate AIDEN features programmatically

Getting Started

1. Launch AIDEN with Remote Debugging

Start AIDEN with the --remote-debugging-port flag to enable CDP:

# macOS
open -a "AIDEN" --args --remote-debugging-port=9222
 
# Linux
aiden-v2-desktop --remote-debugging-port=9222
 
# Windows
"AIDEN.exe" --remote-debugging-port=9222

2. Connect with agent-browser

Once AIDEN is running with debugging enabled:

# Install agent-browser
npm install -g @anthropic-ai/agent-browser
 
# Connect to the running AIDEN instance
agent-browser connect 9222

3. Interact with the App

# Take a snapshot of the current UI state (accessibility tree)
agent-browser snapshot -i
 
# Take a screenshot
agent-browser screenshot kanban-board.png
 
# Click on elements
agent-browser act "click on 'New Project' button"
 
# Type text
agent-browser act "type 'My New App' into project name field"

Example Workflows

Automated Project Setup

#!/bin/bash
# Create a new project via automation
 
# Connect to AIDEN
agent-browser connect 9222
 
# Navigate to project creation
agent-browser act "click on 'New Project'"
agent-browser act "type 'My Automated Project' into the project name input"
agent-browser act "select 'NEXTJS' from project type dropdown"
agent-browser act "click on 'Create Project' button"
 
# Wait for analysis to complete
agent-browser act "wait for 'Analysis Complete' text to appear"
 
# Take a screenshot of the result
agent-browser screenshot project-analysis.png

Automated Screenshot Generation

#!/bin/bash
# Generate documentation screenshots
# See scripts/generate-screenshots.sh for the full version
 
agent-browser connect 9222
OUTDIR=public/screenshots
 
# Capture key screens
agent-browser screenshot "$OUTDIR/dashboard.png"
agent-browser act "click on first project"
agent-browser screenshot "$OUTDIR/kanban-board.png"
agent-browser act "click on Settings"
agent-browser screenshot "$OUTDIR/settings.png"

Story Status Monitoring

#!/bin/bash
# Check on agent progress
 
agent-browser connect 9222
agent-browser snapshot -i | grep -i "in progress"
agent-browser screenshot "agent-progress-$(date +%s).png"

Tips

  • Snapshot vs Screenshot: Use snapshot -i for programmatic inspection (returns accessibility tree text). Use screenshot for visual captures.
  • Waiting: Use agent-browser act "wait for ..." to wait for UI state changes before continuing.
  • Element references: After a snapshot, you can reference elements by their role and name (e.g., button 'Start', textbox 'Project name').
  • Headless mode: For CI environments, combine with xvfb-run on Linux to run without a display.
License & PricingLicense API