Skip to main content

Troubleshooting Guide

This guide helps you identify and resolve the most common issues encountered while using Runable. Whether you’re publishing a website, generating code, creating videos, or working with AI agents, start here before contacting support.

Before You Begin

Before troubleshooting: ✅ Refresh the page ✅ Check your internet connection ✅ Save your latest work ✅ Review any error messages ✅ Retry the task once If the issue persists, continue below.

Let the AI Help You Troubleshoot First

Troubleshooting Debug Workflow
Before contacting support, we recommend asking the AI to diagnose and fix the issue first. Many common problems can be resolved simply by providing the AI with the exact error message instead of describing the issue generally.

Instead of

Try
Providing the complete error message helps the AI identify the root cause much more accurately than a general description.

Ask the AI to Investigate Before Making Changes

Instead of asking the AI to immediately rewrite or regenerate code, ask it to investigate first. Example:
This often produces more focused and reliable results.

If the Chat Stops Responding

Screenshot From 2026 08 06 18 22 57
Occasionally, a conversation may stop responding or become difficult to continue. The recommended recovery method is:
  1. Roll back to the last point where the chat was working correctly.
  2. Continue the conversation from that point.
  3. Reapply your recent changes using smaller, more focused prompts.
Rolling back restores the conversation to the selected point. Any messages and generated changes created after that point will no longer be available, so make sure you no longer need them before rolling back.

Before Contacting Support

If the AI is unable to resolve the issue, our support team will be happy to help. To investigate efficiently, please include:

Shared Chat Link

Share your Runable chat in Shared Mode so our team can review the full project context.

Error Message

Include the complete error message or stack trace exactly as shown.

Screenshots

Attach screenshots of the error or unexpected behavior.

Steps to Reproduce

Explain what you were doing when the issue occurred and how we can reproduce it.

Expected Result

Describe what you expected to happen.

Actual Result

Describe what actually happened.

Support Checklist

Before opening a support request, please confirm:
  • I provided the complete error message to the AI.
  • I asked the AI to investigate before making changes.
  • I tried the suggested fix.
  • I reviewed the latest generated output.
  • I rolled back if the conversation became stuck.
  • I generated a Shared Mode chat link.
  • I have screenshots or error messages ready.
Following these steps helps our support team diagnose issues more quickly and reduces the time needed to reach a resolution.

Website Won’t Publish

Possible Causes

  • Build failed
  • Missing dependency
  • Invalid code
  • Unsupported package
  • Health Check failed
  • Build timeout

How to Fix

  1. Open Preview.
  2. Review build errors.
  3. Fix reported issues.
  4. Publish again.
Large AI-generated code changes may occasionally introduce build issues that require review before publishing.

Health Check Failed

Health Checks verify that your project is ready for deployment. Common causes:
  • Missing imports
  • Broken routing
  • Dependency conflicts
  • Invalid project structure
  • Build errors
  • Review recent changes.
  • Revert large modifications if necessary.
  • Fix reported errors.
  • Republish.

HTTP 500 Error

A 500 Internal Server Error usually indicates that something inside your application failed during execution. Possible causes include:
  • Application code errors
  • Backend exceptions
  • Invalid API configuration
  • Environment variables
  • Database connection issues
  • Third-party integrations
  • Check browser console.
  • Review server logs.
  • Verify API configuration.
  • Verify environment variables.
  • Retry after fixing the underlying issue.

Deployment Failed

Deployment failures may occur because of:
  • Failed Health Check
  • Missing dependencies
  • Build timeout
  • Unsupported configuration
  • Runtime errors
  • Ensure Preview builds successfully.
  • Resolve lint errors.
  • Verify dependencies.
  • Publish again.

AI Didn’t Complete My Task

Sometimes AI generation may stop before completing the requested work. Possible reasons:
  • Task complexity
  • Insufficient available credits
  • Very large project
  • Multiple files requiring modification
  • Break the task into smaller requests.
  • Continue from the previous generation.
  • Improve prompt specificity.

AI Changed Unrelated Files

Large prompts can sometimes affect unrelated parts of your project. Instead of:
Try:

AI Generated Incorrect Results

If the output isn’t what you expected:
  • Add more context.
  • Include examples.
  • Upload references.
  • Specify constraints.
  • Refine your prompt instead of starting over.

Chat Is Stuck

If the AI stops responding: Try:
  • Refreshing the page.
  • Sending another message.
  • Waiting a few moments.
  • Starting a new conversation if necessary.
If the issue persists, contact support.

Credits Were Consumed But Task Didn’t Finish

Very large or resource-intensive tasks may require more credits than are currently available. When credits are exhausted:
  • The AI may stop.
  • Partial work may still be available.
  • Continue from the last completed step after adding more credits or simplifying the task.

Images Don’t Look Correct

Try:
  • Better prompt
  • Reference images
  • Specific art style
  • Camera angle
  • Lighting
  • Aspect ratio

Video Doesn’t Look Correct

Current AI video models work best with:
  • Short text
  • Clear instructions
  • Simple layouts
  • Readable subtitles
Avoid long paragraphs of on-screen text.

Website Looks Different After Publishing

Possible causes:
  • Browser cache
  • CDN propagation
  • Recent deployment
Try:
  • Hard refresh
  • Open in Incognito
  • Wait a few minutes
  • Publish again

Custom Domain Isn’t Working

Verify:
  • DNS records
  • Root domain
  • SSL provisioning
  • Domain propagation
DNS changes may take several hours to fully propagate.

Build Suddenly Broke

Large AI-generated changes may occasionally affect project structure. If this happens:
  • Identify the recent change.
  • Restore a previous working version if available.
  • Apply changes incrementally.

Performance Is Slow

Large applications naturally require more processing. Improve performance by:
  • Splitting large tasks
  • Removing unused code
  • Optimizing assets
  • Using smaller prompts

Still Need Help?

Before contacting support, please include: ✅ Project URL ✅ Screenshots ✅ Error messages ✅ Browser ✅ Steps to reproduce ✅ Prompt used This helps our team investigate your issue more efficiently.

Related Guides

  • Prompt Engineering Guide
  • AI Limitations
  • Understanding Credit Usage
  • Website Hosting
  • Billing Guide