> ## Documentation Index
> Fetch the complete documentation index at: https://docs.bugster.dev/llms.txt
> Use this file to discover all available pages before exploring further.

# Destructive

> Run AI-powered destructive agents to find potential bugs in your changed pages

The `bugster destructive` command runs AI-powered destructive agents to find potential bugs by simulating edge cases, unexpected user behavior, and malicious inputs that could break your application.

## Basic Usage

```bash theme={null}
bugster destructive
```

## Command Overview

Executes specialized AI agents designed to stress-test your application by attempting to break it through aggressive interactions, malicious inputs, and edge case scenarios.

### When to Use

* **Before major releases** to catch edge case bugs
* On staging environments to stress-test changes
* After significant code modifications
* When you want to find vulnerabilities proactively

### Key Features

* AI agents simulate malicious and edge case user behavior
* Only targets pages with recent code changes (git diff)
* Records videos of destructive attempts for debugging
* Multiple specialized agent types with different attack vectors
* More aggressive than regular testing approaches

<Warning>
  **NEVER run against production environments.** Destructive agents can cause data corruption, submit forms, delete data, or trigger irreversible actions. Use only on test/staging environments.
</Warning>

## Command Syntax

```bash theme={null}
bugster destructive [options]
```

## How It Works

The destructive command:

1. **Change Detection**: Identifies pages with recent code changes using git diff
2. **Agent Selection**: Deploys specialized destructive agents based on page types
3. **Aggressive Testing**: Agents attempt to break functionality through various methods
4. **Result Recording**: Captures videos and logs of all destructive attempts
5. **Bug Analysis**: Reports potential vulnerabilities and edge case failures

## Agent Types

<Tabs>
  <Tab title="UI Crashers">
    **Priority**: Highest

    Attempts to crash UI components through:

    * Rapid clicking and interaction spam
    * Invalid input combinations
    * Boundary value testing
    * Component state corruption
  </Tab>

  <Tab title="Form Destroyer">
    **Priority**: High

    Targets forms with malicious inputs:

    * SQL injection attempts
    * XSS payload testing
    * Buffer overflow attempts
    * Invalid data type submissions
  </Tab>
</Tabs>

## Execution Options

<ParamField query="--headless" type="boolean">
  Run agents in headless mode without visible browser UI. Recommended for CI/CD environments.
</ParamField>

<ParamField query="--parallel" type="integer" default="3">
  Maximum number of concurrent agents to run simultaneously. Higher values increase testing speed but use more resources.
</ParamField>

<ParamField query="--max-concurrent" type="integer" default="3">
  Alias for `--parallel`. Specifies the maximum number of agents to run concurrently.
</ParamField>

<ParamField query="--limit" type="integer" default="5">
  Maximum number of destructive agents to execute. Range: 1-30. Higher values increase bug discovery chances.
</ParamField>

<ParamField query="--browser" type="string" default="chromium">
  Browser engine to use for destructive testing. Supports chromium, firefox, and webkit.
</ParamField>

## Output Options

<ParamField query="--stream-results" type="boolean" default="true">
  Stream real-time results to the Bugster dashboard as agents complete their destruction attempts.
</ParamField>

<ParamField query="--no-stream-results" type="boolean">
  Disable streaming to dashboard. Results will only be available locally.
</ParamField>

<ParamField query="--base-url" type="string">
  Override the base URL from configuration. Essential for testing staging environments.
</ParamField>

<ParamField query="--verbose" type="boolean">
  Enable detailed logging of agent execution and attack patterns.
</ParamField>

<ParamField query="--silent" type="boolean">
  Run agents silently with minimal output. Reduces noise in automated environments.
</ParamField>

## Examples

### Complete Destructive Testing

```bash theme={null}
bugster destructive
```

Runs all available destructive agents on pages with recent changes.

### Pre-release Validation

```bash theme={null}
bugster destructive --limit 15 --parallel 5
```

Comprehensive destructive testing with maximum coverage before major releases.

### Quick Vulnerability Scan

```bash theme={null}
bugster destructive --limit 3 --headless
```

Fast security and stability check with minimal resource usage.

### Staging Environment Testing

```bash theme={null}
bugster destructive --base-url https://staging.myapp.com --limit 10
```

Test staging environment with comprehensive destructive scenarios.

### CI/CD Integration

```bash theme={null}
bugster destructive --headless --stream-results --silent
```

Automated destructive testing optimized for continuous integration pipelines.

## Usage Patterns

<CardGroup cols={2}>
  <Card title="Pre-release Testing" icon="rocket">
    ```bash theme={null}
    bugster destructive --limit 15
    ```

    Comprehensive destructive testing before major releases.
  </Card>

  <Card title="Quick Security Check" icon="shield">
    ```bash theme={null}
    bugster destructive --limit 3 --headless
    ```

    Fast vulnerability and stability validation.
  </Card>

  <Card title="CI/CD Integration" icon="gear">
    ```bash theme={null}
    bugster destructive --headless --stream-results
    ```

    Automated testing in continuous integration pipelines.
  </Card>

  <Card title="Performance Testing" icon="bolt">
    ```bash theme={null}
    bugster destructive --parallel 8
    ```

    Maximum concurrent destruction for stress testing.
  </Card>
</CardGroup>

## Safety Guidelines

<AccordionGroup>
  <Accordion title="Environment Safety">
    * **Never run against production environments**
    * Use dedicated test databases and isolated environments
    * Ensure all data is disposable or properly backed up
    * Test on staging environments that mirror production
  </Accordion>

  <Accordion title="Data Protection">
    * Agents may submit forms with test data
    * Potential for data deletion or corruption
    * May trigger irreversible actions
    * Always use test credentials and mock data
  </Accordion>

  <Accordion title="Video Evidence">
    * All destructive attempts are automatically recorded
    * Videos help understand what caused detected issues
    * Review recordings to learn about attack patterns
    * Use videos for security team analysis
  </Accordion>
</AccordionGroup>

## Result Analysis

After execution, you'll receive:

* **Vulnerability Reports**: Detailed analysis of potential security issues
* **Edge Case Failures**: Documentation of unexpected behavior patterns
* **Video Recordings**: Visual evidence of successful attacks or crashes
* **Agent Logs**: Detailed execution traces for debugging
* **Severity Ratings**: Prioritized list of discovered issues

## Best Practices

<Tip>
  Run destructive agents after `bugster update` to test newly modified pages. Higher `--limit` values increase the chance of finding edge case bugs, and videos are automatically saved for any issues found.
</Tip>

<Note>
  Destructive testing is most effective as a final validation step before releases. Combine with regular testing for comprehensive coverage, and always review video recordings to understand root causes.
</Note>