Troubleshooting

Generation fails with 500

• Check backend logs for Ollama connection errors.
• Verify API key and base URL when using cloud models.

Execution errors

• fetch failed: network reset when calling the model provider.
• max iterations reached: lower max steps or provide clearer assertions.
• timeout: reduce page load strategy or run in fast mode.

Tips

• Use stable URLs and deterministic UI states.
• Prefer smaller prompts with explicit expectations.
• Use repo context to improve selector quality.