benbrown

This is a transcript of a machine-generated story
that took place in fiction.quest. The lines starting
with > were my commands. The rest are the hallucinations
of a machine that was raised on our popular culture.

https://fiction.quest

> Rufus, let's take Max on a time travel adventure

Rufus grins widely at Zeb, his eyes lighting up with 

benbrown

// install these libs, set OPENAI_API_KEY, and put your baseline prompt in prompt.txt.

require('dotenv').config()
const { Configuration, OpenAIApi } = require("openai");
const wordwrapjs = require('wordwrapjs');
const fs = require('fs');
const prompt = require('prompt-sync')({sigint: true});

const configuration = new Configuration({
  apiKey: process.env.OPENAI_API_KEY,

benbrown

How to use multi-bot template:

Get the yeoman generator

• Checkout the components repo: https://github.com/microsoft/botframework-components

Run npm install inside the generator packages

benbrown

[
   {
      "name": "benbrown/grammarHelperLG",
      "version": "",
      "source": "npm",
      "language" :"js",
      "repository": "https://github.com/benbrown/grammarHelperLG",
      "authors": ["Ben Brown <benbrown@gmail.com>"],
      "category": "Community Packages",
      "description": "A collection of LG rules to help form proper sentences out of dynamic data"

benbrown

How to build a bot framework compatible adapter

Bot Framework uses a concept called "adapters" to create an interface between external messaging platforms and the bot logic. Adapters are responsible for passing messages and events back and forth between the bot and users, and translating them between their external and internal forms.

There are many open source adapters available (for Node, for C#), but there are many messaging platforms still unsupported. The code for these adapters provides a good template for creating a new one.

The Azure Bot Service channel system can be thought of as a meta-adapter: in this case, the translation is done in the cloud by the service, and

What does an adapter do?

benbrown

Switch npm to pull from the Bot Framework nightly build to install botbuilder-lg's preview release:

npm config set registry https://botbuilder.myget.org/F/botbuilder-declarative/npm/
npm install --save botbuilder-lg@4.5.0

Put this code in a Botkit skill:

benbrown

/**
* This module if installed into a botkit v4+ app will enable "Bot Inspector" mode in Bot Framework Emulator.
* Put it in features/botinspector.js inside your fresh Botkit app.
* This allows you to connect to the bot app running locally with emulator to inspect messages 
* as they come and go to the messaging platform.
* Read more here:
* https://github.com/Microsoft/botframework/blob/master/README.md#bot-inspector-new---preview
*/
// import botbuilder 4.4 library
const { InspectionMiddleware, InspectionState, BotFrameworkAdapter } = require('botbuilder')

benbrown

var schedule = require('node-schedule');

// channel/group ID fields for the places you want to send messages
var groupChannels = ['G6BKKTVPG', 'C6WHTV57V'];

// your slack team's id, needed to spawn a bot
var team_id = 'T123123';

// this can probably be a FAKE user id, just required as part of startConversation
var user_id = 'U123'; 

benbrown

Botkit.on('connected', function() {

  Botkit.triggerScript('script','thread');
  
});

benbrown

var user_id = 123;

app.controller('chat', ['$scope',function($scope, $cookies) {

  var socket;

  $scope.reconnect_count = 0;
  $scope.reconnect_timeout = 3000;
  $scope.max_reconnect = 5;