the-rabak/reproduce-bug

Goal

Investigate and reproduce a reported bug with parallel research, browser evidence when relevant, and a documented issue response.

Use this skill when

• A GitHub issue or bug report needs verification and investigation.
• Reproduction steps, screenshots, code-path findings, and suggested fixes are needed.
• UI-related bugs may need browser automation evidence.

Do not use this skill when

• The task is a simple implementation bug fix rather than investigation.
• The issue cannot be accessed and no usable description is available.
• Browser evidence is impossible and the bug is purely non-visual.

Non-negotiable rules

• Read the issue description and comments first.
• Phase 1 research must run repo-research-analyst and learnings-researcher in parallel.
• Repeat the research loop until you have a solid hypothesis about where the bug could be coming from.
• If the bug is UI-related, verify the local server before browser reproduction.
• Capture screenshots as evidence whenever you reproduce the bug visually.
• Document findings with concrete file paths and, when possible, line references.
• Report back in a structure that can be posted to the issue.

Procedure

Execute the full source-of-truth workflow below. Preserve every conditional branch, phase, checklist, and validation step from the original instructions. When the workflow says to ask, wait, route, or run in parallel, do that exactly.

Arguments

[issue number or description]

Reproduce Bug Command

Look at github issue #$ARGUMENTS and read the issue description and comments.

Phase 1: Log Investigation

Run the following agents in parallel to investigate the bug:

1. Task repo-research-analyst(issue_description) - Search codebase for relevant code paths
2. Task learnings-researcher(issue_description) - Check if similar bugs have been solved before

Think about the places it could go wrong looking at the codebase. Look for logging output we can look for.

Run the agents again to find any logs that could help us reproduce the bug.

Keep running these agents until you have a good idea of what is going on.

Phase 2: Visual Reproduction with agent-browser

If the bug is UI-related or involves user flows, use agent-browser to visually reproduce it:

Step 1: Verify Server is Running

agent-browser open http://localhost-3000
agent-browser snapshot -i

If server not running, inform user to start the development server for the project.

Step 2: Navigate to Affected Area

Based on the issue description, navigate to the relevant page:

agent-browser open "http://localhost:3000/[affected_route]"
agent-browser snapshot -i

Step 3: Capture Screenshots

Take screenshots at each step of reproducing the bug:

agent-browser screenshot "bug-[issue]-step-1.png"

Step 4: Follow User Flow

Reproduce the exact steps from the issue:

1. Read the issue's reproduction steps

2. Execute each step using agent-browser:

   • agent-browser click @ref for clicking elements
   • agent-browser fill @ref "text" for filling forms
   • agent-browser snapshot -i to see the current state
   • agent-browser screenshot filename.png to capture evidence

3. Check for console errors by inspecting the snapshot output for error indicators.

Step 5: Capture Bug State

When you reproduce the bug:

1. Take a screenshot of the bug state
2. Document any console errors
3. Document the exact steps that triggered it

agent-browser screenshot "bug-[issue]-reproduced.png"

Phase 3: Document Findings

Reference Collection:

• [ ] Document all research findings with specific file paths (e.g., app/Services/ExampleService.php:42)
• [ ] Include screenshots showing the bug reproduction
• [ ] List console errors if any
• [ ] Document the exact reproduction steps

Phase 4: Report Back

Add a comment to the issue with:

1. Findings - What you discovered about the cause
2. Reproduction Steps - Exact steps to reproduce (verified)
3. Screenshots - Visual evidence of the bug (upload captured screenshots)
4. Relevant Code - File paths and line numbers
5. Suggested Fix - If you have one

Required output

Return an issue-comment-ready report with:

• Findings.
• Verified reproduction steps.
• Screenshot evidence.
• Relevant code paths.
• Suggested fix, if available.