cleverbot (work in progress)

We have a knowledgebase that’s a Jekyll website. This bot gives us a slash command in Slack that enables us to search that site.

Installation

(For now)

1. Clone the repository
2. bundle install

Indexing your site

Run the following command either in a Git hook or a cron job. It will build a persistent index of the content on your site:

$ bin/index path/to/jekyll/site

The index command assumes that posts are stored in a directory called _posts, one level below the path that you pass to the command.

Setting up Slack

You need to create two integrations in Slack:

1. A slash command that points to the URL of your server
2. An incoming webhook that the search server will use to post search results into Slack

Running the server

The search server is a Sinatra application. It relies on a few environment variables:

• SOURCE_DIRECTORY — the fully qualified path to your Jekyll site.
• SLACK_TOKEN — the token for your slash command in Slack
• SLACK_WEBHOOK_URL — the URL of your incoming webhook integration in Slack
• SITE_URL — the URL of the homepage of your site, used as the base for article URLs in search results

With these environment variables set, use rackup to start the server:

$ SOURCE_DIRECTORY=”/var/www/kb” SLACK_TOKEN=”abc123abc123” SLACK_WEBHOOK_URL=”https://hooks.slack.com/services/foo/bar/baz” SITE_URL=”http://example.com” rackup

It’s assumed that your service will be very low traffic, and therefore that running the server through WEBrick is fine.