Skip to main content

Creating Browser Agents

Browser agents automate web interactions to test workflows, monitor applications, and validate functionality. This guide walks you through creating and configuring your first browser agent.

  1. Log in to your DebuggAI dashboard
  2. Click Browser Agents in the left navigation menu
  3. You'll see a list of existing agents (if any) and the option to create new ones

Create New Agent​

Click the Create Agent button to open the agent creation form.

Required Fields​

FieldDescription
NameA descriptive name for the task (e.g., "Checkout Flow Test", "Login Validation")
DescriptionWhat the agent should accomplish. Be specific about the steps and expected outcomes
ProjectSelect the associated project from your workspace

Optional Fields​

FieldDescription
TagsAdd tags for organization and filtering (e.g., "checkout", "critical-path", "staging")

Writing Effective Descriptions​

Your description tells the agent what to do. Write clear, step-by-step instructions:

1. Navigate to the login page
2. Enter test credentials (user@example.com / testpass123)
3. Click the login button
4. Verify the dashboard loads successfully
5. Check that the user's name appears in the header

Configure Agent Settings​

After creating the agent, configure its execution settings.

Target URL/Environment​

Specify where the agent should run:

  • Development: https://dev.yourapp.com
  • Staging: https://staging.yourapp.com
  • Production: https://yourapp.com

Authentication Requirements​

If your agent needs to access authenticated areas:

  • Provide test credentials in the agent description
  • Use environment-specific test accounts
  • Avoid using production user credentials

Success Criteria​

Define what constitutes a successful run:

  • Page elements that should be visible
  • URLs the agent should reach
  • Text content that should appear
  • Actions that should complete without errors

Initial Setup Best Practices​

Start Simple​

Begin with focused, single-purpose agents:

  • Test one user flow per agent
  • Keep the number of steps manageable (5-10 steps)
  • Avoid complex conditional logic initially

Use Clear Descriptions​

Good descriptions lead to reliable agents:

Good:

Navigate to /products, click the first product card,
verify the product detail page loads with a price displayed

Avoid:

Test the products

Test in Development First​

  1. Create agents targeting your development environment
  2. Run several times to verify consistency
  3. Adjust descriptions based on results
  4. Promote to staging/production once stable

Agent Metadata​

Tracking Runs​

Each agent tracks execution history:

  • Run Count: Total number of executions
  • Success Rate: Percentage of successful runs
  • Last Run: Timestamp and status of most recent execution
  • Average Duration: Typical execution time

Version History​

Agents maintain a history of changes:

  • Description modifications
  • Configuration updates
  • Tag changes

Review version history to understand how an agent has evolved and revert if needed.

Next Steps​

Once your agent is created:

  1. Run manually to verify it works as expected
  2. Review the execution log for any issues
  3. Schedule recurring runs for continuous monitoring
  4. Set up alerts to notify you of failures

Next: Learn about Training Agents to improve accuracy.