OpenAssistant

Deploy

• python3 -m venv venv && ./venv/bin/pip install -r requirements-server.txt
• cp settings-example.toml settings.toml
• $EDITOR settings.toml
• $EDITOR openas-backend.service
• Profit

Request Spec

{
    "mysecret": "SOMETHING", // matching one of ws valid_tokens in settings.toml
    "thread_id": "SOME_STRING", // or 'NA' if new conversation
    "message": "STRING_FROM_USER", // what did user say
    "username": "john doe" // what call the user
}

Response Spec

{
    "response": "THING", // string response from OpenAI or error msg
    "thread_id": "STRING", // prev. thread ID or new ID if request was 'NA'
    "username": "STRING" // echos value of request
}

Suggestion

If you're testing server and one or more clients on the same machine, just ln -s settings.toml ../settings.toml