scoutqa-test

ScoutQA Testing Skill

Perform AI-powered exploratory testing on web applications using the scoutqa CLI.

Think of ScoutQA as an intelligent testing partner that can autonomously explore, discover issues, and verify features. Delegate testing to multiple parallel ScoutQA executions to maximize coverage while saving time.

When to Use This Skill

Use this skill in two scenarios:

1. User requests testing - When the user explicitly asks to test a website or verify functionality
2. Proactive verification - After implementing web features, automatically run tests to verify the implementation works correctly

Example proactive usage:

• After implementing a login form → Test the authentication flow
• After adding form validation → Verify validation rules and error handling
• After building a checkout flow → Test the end-to-end purchase process
• After fixing a bug → Verify the fix works and didn't break other features

Best practice: When you finish implementing a web feature, proactively start a ScoutQA test in the background to verify it works while you continue with other tasks.

Running Tests

Testing Workflow

Copy this checklist and track your progress:

Testing Progress:

• Write specific test prompt with clear expectations
• Run scoutqa command in background
• Inform user of execution ID and browser URL
• Extract and analyze results

Step 1: Write specific test prompt

See "Writing Effective Prompts" section below for guidelines.

Step 2: Run scoutqa command

IMPORTANT: Use the Bash tool's timeout parameter (5000ms = 5 seconds) to capture execution details:

When calling the Bash tool, set timeout: 5000 as a parameter:

• This is the Bash tool's built-in timeout parameter in Claude Code (NOT the Unix timeout command)
• After 5 seconds, the Bash tool returns control with a task ID and the process continues running in the background
• This is different from Unix timeout which kills the process - here the process keeps running
• The first 5 seconds capture the execution ID and browser URL from ScoutQA's output
• The test continues running remotely on ScoutQA's infrastructure with the background task

scoutqa --url "https://example.com" --prompt "Your test instructions"

In the first few seconds, the command will output:

• Execution ID (e.g., 019b831d-xxx)
• Browser URL (e.g., https://app.scoutqa.ai/t/019b831d-xxx)
• Initial tool calls showing test progress

After the 5-second timeout, the Bash tool returns a task ID and the command continues running in the background. You can work on other tasks while the test runs. The timeout is only to capture the initial output (execution ID and browser URL) - the test keeps running both locally as a background task and remotely on ScoutQA's infrastructure.

Step 3: Inform user of execution ID and browser URL

After the Bash tool returns with the task ID (having captured the execution details in the first 5 seconds), inform the user of:

• The ScoutQA execution ID and browser URL so they can monitor progress in their browser
• The background task ID if they want to check local command output later

The test continues running in the background while you continue other work.

Step 4: Extract and analyze results

See "Presenting Results" section below for the complete format.

Command Options

• --url (required): Website URL to test (supports localhost / 127.0.0.1)
• --prompt (required): Natural language testing instructions
• --project-id (optional): Associate with a project for tracking
• -v, --verbose (optional): Show all tool calls including internal ones

Local Testing Support

ScoutQA supports testing localhost and 127.0.0.1 URLs autonomously — no manual setup required.

# Seamlessly test a locally running app when you're developing your app
scoutqa --url "http://localhost:3000" --prompt "Test the registration form"

When to Use Each Command

Starting a new test? → Use scoutqa --url --prompt Verifying a known issue? → Use scoutqa issue-verify --issue-id <id> Finding issue IDs from an execution? → Use scoutqa list-issues --execution-id <id> Agent needs more context? → Use scoutqa send-message (see "Following Up on Stuck Executions")

Writing Effective Prompts

Focus on what to explore and verify, not prescriptive steps. ScoutQA autonomously determines how to test.

Example: User registration flow

scoutqa --url "https://example.com" --prompt "
Explore the user registration flow. Test form validation edge cases,

Example: E-commerce checkout

Example: Running parallel tests for comprehensive coverage

Launch multiple tests in parallel by making multiple Bash tool calls in a single message, each with the Bash tool's timeout parameter set to 5000 (milliseconds):

Implementation: Send a single message with three Bash tool calls. For each Bash tool invocation, set the timeout parameter to 5000 milliseconds. After 5 seconds, each Bash call returns with a task ID while the processes continue running in the background. This captures the execution ID and browser URL from each test in the initial output, then all three continue running in parallel (both as background tasks locally and remotely on ScoutQA's infrastructure).

Key guidelines:

• Describe what to test, not how to test (ScoutQA figures out the steps)
• Focus on goals, edge cases, and concerns
• Run multiple parallel executions for different test areas
• Trust ScoutQA to autonomously explore and discover issues
• Always set the Bash tool's timeout parameter to 5000 milliseconds when calling scoutqa commands (this returns control after 5 seconds while the process continues in the background)
• For parallel tests, make multiple Bash tool calls in a single message
• Remember: Bash tool timeout ≠ Unix timeout command (Bash timeout continues the process in background, Unix timeout kills it)

Common Test Scenarios

Post-deployment smoke test:

Accessibility audit:

E-commerce testing:

SaaS application:

Form validation:

Mobile responsiveness:

Verification of a known issue:

The issue-verify command will:

1. Create a verification execution for the issue
2. Show the execution ID and browser URL
3. Stream the agent's verification progress in real-time
4. Display a completion summary with a link to results

Feature verification (after implementation):

Example: Proactive testing after coding a feature

After implementing a user registration form, automatically verify it works:

This catches issues immediately while the implementation is fresh in context.

Listing Issues

Use scoutqa list-issues to browse issues found in a previous execution. This is useful for finding issue IDs to use with issue-verify.

Options:

• --execution-id (required): Execution ID (from the /t/<executionId> URL or CLI output)

Example output:

Presenting Results

Immediate Presentation (After Starting Test)

Right after running the scoutqa command, present the execution details to the user:

Final Results (After Completion)

When the execution completes, use this format to present findings:

Always include:

• Execution ID (e.g., ex_abc123) for reference
• Issues found with severity, category (accessibility, usability, functional), impact, and location

Following Up on Stuck Executions

If the remote agent gets stuck or needs clarification, use send-message to continue:

Checking Test Results

ScoutQA tests run remotely on ScoutQA's infrastructure. After starting a test with a short timeout to capture the execution ID:

1. The test continues running remotely (not locally in background)
2. You can continue other work immediately
3. To check results later, visit the browser URL provided when the test started
4. Alternatively, use scoutqa get-execution --execution-id <id> to fetch results via CLI

Best practice: Start tests by setting the Bash tool's timeout parameter to 5000 milliseconds. After 5 seconds, the Bash tool returns control with a task ID and the execution details (execution ID and browser URL) while the test continues running in the background. You can then continue other work and check results on ScoutQA's website or via CLI when needed.

Troubleshooting