The bugster destructive command runs specialized AI agents that actively explore your application to find potential bugs, edge cases, and unexpected behaviors in pages that have been modified.

Run bugster init first if you haven’t already set up your project.

Basic Usage

bugster destructive

This command will:

  1. Analyze git changes to identify modified pages
  2. Assign specialized agents to each changed page
  3. Run agents concurrently to explore and test your application
  4. Record interactions with video capture for debugging
  5. Report discovered bugs with detailed descriptions

How It Works

1

Change Detection

Analyzes git changes against origin/main to identify modified Next.js pages and their dependencies.

2

Agent Assignment

AI determines which specialized agents should test each page based on detected changes.

3

Destructive Exploration

Agents autonomously explore pages, testing edge cases and recording all interactions.

4

Bug Detection

Reports JavaScript errors, UI issues, failed requests, and performance problems.

Command Options

Basic Options

bugster destructive --headless --max-concurrent 5 --verbose

Results

Output Table

┌──────────────────────────────────────────────────────────────┐
│                    Destructive Agent Results                 │
├─────────────────────┬─────────────────┬─────────────┬────────┤
│ Page                │ Agent           │ Bugs Found  │ Time   │
├─────────────────────┼─────────────────┼─────────────┼────────┤
│ pages/auth/login    │ Agent_1         │ 2           │ 45.2s  │
│ pages/dashboard     │ Agent_2         │ 0           │ 32.1s  │
└─────────────────────┴─────────────────┴─────────────┴────────┘

Video Recordings

Agent interactions are recorded and saved to:

.bugster/videos/destructive/{run-id}/{agent}_{page}/

Exit Codes

  • Exit Code 0: No bugs found
  • Exit Code 1: Bugs discovered - review report and videos

Best Practices

  • Start with low concurrency (--max-concurrent 1-3) for shared environments
  • Use dedicated test environments - avoid production
  • Review video recordings to understand agent behavior
  • Validate reported bugs - some may be environment-specific

Troubleshooting

Next Steps