Always "actions.intent.MAIN" with Google Conversation Actions?

google-assistant

#1

Hey :slight_smile:

I am developing a skill for Alexa and Google Conversational Actions. With Alexa everything works fine but with GCA i have a big problem (probably something small). Everytime i want to test my skill with the Assisant i get the information that my skill is not repsoning correctly and i sould wait for a few minutes. The server is running and i get the events when i set logging to true. In the jovo debugger i can see the incoming intent aswell. I see the new intent with the name “actions.intent.MAIN” (always) and in the “query” attribute i get what i actually said to the device.

The Problem is now, that it always responses with the “LAUNCH”-Handler and nothing else. It seems like a part of the routing is not working correctly or maybe i missunderstand something.

(Maybe Second Problem) I tried to give an answer with tell in de LAUNCH-Handler to verify if anything works. That worked, but adding a mediaresponse lets google response with the same text like before. The Skill is not responding and i should try it later. Looking at the response generated by jovo i got a valid looking response with my livestream in the media section and all text and images…

I did not remap the “actions.intent.MAIN” and i deploy my langauge model through Jovo. I am using the package “jovo-platform-googleassistantconv” in the Version 3.4.1.

Someone has an idea what could the problem be? Or is there a missunderstanding on my side?

Thanks for your time! :slightly_smiling_face:

EDIT
I migrated the old Dialogflow Project to Google Conversational Actions


#2

Could you provide a sample request payload?


#3

I got some new findings.

The language model on google side seems to be working. (partially)
Intents from the Dialogflow-Era still work and get routed without problems. I even updated the naming of some intents and the routing still worked fine. But new intents dont get routed.

This is the raw request from the debugger:

{
  handler: { name: 'Jovo' },
  intent: {
    name: 'actions.intent.MAIN',
    params: {},
    query: 'spreche mit voice experiment und starte gurkensalat'
  },
  scene: {
    name: 'actions.scene.START_CONVERSATION',
    slotFillingStatus: 'UNSPECIFIED',
    slots: {}
  },
  session: {
    id: '...',
    params: {},
    typeOverrides: [],
    languageCode: ''
  },
  user: {
    locale: 'de-DE',
    params: { userId: '...' },
    accountLinkingStatus: 'ACCOUNT_LINKING_STATUS_UNSPECIFIED',
    verificationStatus: 'VERIFIED',
    packageEntitlements: [],
    gaiamint: '',
    permissions: [],
    lastSeenTime: '2021-02-24T11:03:54Z'
  },
  home: { params: {} },
  device: { capabilities: [ 'SPEECH', 'RICH_RESPONSE', 'LONG_FORM_AUDIO' ] }
}

Creating a new google conversational project does not change the situation. So i think it doesnt have to do anything with google at that point. The requested intentname has no numbers or anything special in it (MEDIA_STREAM_GURKENSALAT_LIVE).

Just for testing i also tried if any function creates a problem. When i used the LAUNCH handler i was able to respond with text. But adding a mediaresponse lets google respond with “the actions is not responding …”. Maybe it is a corresponding error or an other problem. I am not sure.

Thx for your help :slight_smile: