Api.ai Slack Integration

Overview

Api.ai Slack integration allows you to create Slack bots with natural language understanding based on Api.ai technology.

To launch a bot, you’ll need the Linux OS. To launch it in other operating systems, use Docker Toolbox.

Api.ai documentation:

• How to create an Api.ai agent
• How to obtain Api.ai authentication keys

You’ll need 2 keys:

• Client access token for Api.ai
• Slack bot API token

To obtain a Slack bot API token, create a new bot integration here: https://slack.com/apps/A0F7YS25R-bots.

Bot Launch

To launch the bot, use one of the following commands:

For background launch mode (-d parameter):

docker run -d --name slack_bot \
           -e accesstoken="api.ai access key" \
           -e slackkey="slack bot key" \
           speaktoit/api-ai-slack-bot

For interactive launch mode (-it parameter):

docker run -it --name slack_bot \
           -e accesstoken="Api.ai client access key" \
           -e slackkey="Slack bot user key" \
           speaktoit/api-ai-slack-bot

To stop the bot from running in the interactive mode, press CTRL+C.

In the background mode, you can control the bot’s state via simple commands:

• docker start slack_bot
• docker stop slack_bot,

where slack_bot is the container name from the run command.

Custom Bot Launch

If you want to customize your bot behavior, follow the steps below.

1. Clone the repository https://github.com/api-ai/api-ai-slack-bot

2. Change the code to index.js

3. In the Docker, use the run command specifying the full path to the directory containing the index.js file:

docker run -d --name slack_bot \
           -e accesstoken="Api.ai client token" \
           -e slackkey="Slack bot user key" \
           -v /full/path/to/your/src:/usr/app/src \
           speaktoit/api-ai-slack-bot

Code Notes

Bot implementation is based on the Slack Botkit: https://github.com/howdyai/botkit.

Message processing is done by the following code:

controller.hears(['.*'],['direct_message','direct_mention','mention', 'ambient'], function(bot,message) {
    console.log(message.text);
    if (message.type == "message") {
        if (message.user == bot.identity.id) {
            // message from bot can be skipped
        }
        else {
            var requestText = message.text;
            var channel = message.channel;
            if (!(channel in sessionIds)) {
                sessionIds[channel] = uuid.v1();
            }
            var request = apiAiService.textRequest(requestText, { sessionId: sessionIds[channel] });
            request.on('response', function (response) {
                console.log(response);
                if (response.result) {
                    var responseText = response.result.fulfillment.speech;
                    if (responseText) {
                        bot.reply(message, responseText);
                    }
                }
            });
            request.on('error', function (error) {
                console.log(error);
            });
            request.end();
        }
    }
});

This code extracts the text from each message:

var requestText = message.text;

And sends it to Api.ai:

var request = apiAiService.textRequest(requestText, { sessionId: sessionIds[channel] });

If a non-empty response is received from Api.ai, the bot will respond with the received text:

bot.reply(message, responseText);