5 Common Error Messages for Custom Alexa Skills and How to Troubleshoot Them
When developing your first Alexa skill, issues can arise in areas that you don't expect. For developers who are just starting out and trying to test their first Hello World skill, you may not be able to tell if the issue is with your own code or if there's something missing in your configuration.
Luckily, many of the issues that new skill developers face boil down to some simple misconfigurations that are relatively easy to resolve. Below, we outline 5 of the most common issues that developers run into when testing their skills, using the test simulator with the developer console; the reason why they occur; and how to resolve the issue before submitting your skill for certification.
1. Alexa: "There was a problem with the requested skill's response."
Now that you have finished building out a rough skill and you want to test it out, you can test it either in developer console or your Alexa device. When you invoke your skill by saying your launch phrase, you might run into the situation described below.
- Customer: “Alexa, launch/open <skill's invocation name>.”
- Alexa: “There was a problem with the requested skill's response.”
You might also run into this error message while you are testing other intents, not just LaunchRequest as well.
So, what does that error message mean? When you launched your skill, your skill sent back a response to LaunchRequest or another intent request; however, there was an issue with the response your skill sent back which caused the skill to end the session. This means that the response JSON your skill sent back was malformed.
How to Troubleshoot:
In order to take a closer look at your response JSON, be sure to log your response JSON to improve your ability to debug. This helps you to see what your skill is sending back. If you are using an AWS Lambda function as your skill's backend, making use of test event could help.
2. Alexa: "Unable to reach the requested skill"
- Customer: “Alexa, launch/open <skill's invocation name>.”
- Alexa: “Unable to reach the requested skill.”
This response may occur when you use an endpoint that is not an AWS Lambda function. Alexa will try to reach your endpoint; but, due to incorrect SSL settings, Alexa is unable to hit your endpoint and returns this error message.
How to Troubleshoot:
Depending on the SSL certificate that your custom web service is using, you will want to select the relevant setting within your endpoint. Make sure that your endpoint complies with the requirement listed in Host a Custom Skill as a Web Service.
3. Alexa: "Hmm I don't know that"
When you test your skill in the simulator within the developer console, you are able to invoke your skill correctly; however, when you test your skill on a device, you run into this error. This issue usually occurs if you are using Household Profiles on Alexa devices, where the account that is signed into the Alexa device might not be the account that you are using to develop your skill.
How to Troubleshoot:
You can check the profile that you are currently using for the device by asking “Alexa, which account is this?”. If the account mentioned is not the developer account that you are using to create your skill, you can then switch profile to the one that has your skill enabled by saying “Switch accounts”.
4. Alexa: <Audio only> sound or earcon sound or earcon
When you test your newly built skill within the test simulator on the developer console, you might run into the following message:
- Customer: “Alexa, <utterance> ”
- Alexa: “ <error tone> ”
This response occurs when Alexa cannot match what you just said to an utterance in your interaction model. Alexa will try to match what you say to her, but sometimes there are values that just do not match to any valid utterances, resulting in a low confidence response from Alexa. When this happens, Alexa will return the response, signifying that she was not able to confidently match what you just said to what you have defined in your interaction model.
How to troubleshoot:
Use the Utterance Profiler in the developer console to test the different utterances that are not working by examining how your utterances are being mapped to your defined intents. This will let you see what intents Alexa is trying to map your utterance to. It will also give you an idea of which utterances are working better for your skill, as well as which utterances are not correctly entering the intents.
5. Alexa: <No response>
When trying to test your skill within test simulator, you might not receive a response, or your utterance never finishes loading. This is usually due to your browser and environment not being compatible with the Alexa test simulator.
Alternatively, when you are developing a skill for a locale that is different from the locale your developer account is associated with, you might run into this issue. An example is that you are developing a skill that has an en-GB language model, but you are based in the US. During the process of testing your skill, you try to launch your skill on a device. But no response is returned from Alexa. This issue could occur if you have not correctly set the country and region settings.
How to troubleshoot:
- Browser issues
- Use Chrome/Firefox for testing
- Disable any extensions or plugins you may have installed
- Run your browser in private/incognito modes
- Clear your browser cache
- Local environment issues
- Check your antivirus software and firewalls that could potentially block the test simulator and ports accessed by your browser.
- Check your local network settings that could potentially block the test simulator and ports accessed by your browser
- Update your device country settings
Now that you have learned about 5 of the most common error responses that Alexa gives and of what they mean, hopefully debugging your skill is much easier to handle. If you have any questions along your skill building adventure, you can post questions and discuss with other fellow Alexa skill builders on our developer forum.