Instant Wiki

Use NLP to extract an automatic and dynamically updated references and keywords section to the end of any Confluence article.

Built on Atlassian Forge.


How many times have you come across an article that uses terminology or topics you don't understand? Many times you may be writing documentation yourself for items where additional references might be useful. A references section is something also frequently overlooked when writing documentation, but may help unfamiliar colleagues or readers understand your content through outside sources.

What it does

  • Instant Wiki is a macro that automatically adds an appendix section to any Confluence article
  • Uses basic NLP to extract core phrases and topics out of an article. Extracting keywords can help a reader get a quick feel for the gist of a long article, and improve search results.
  • Generates automatic wikipedia links for additional information on any of the extracted key phrases from the article.

Search for the instant-wiki forge app

Invoke via /Create wiki

How I built it

  • Macro parses out the paragraph (body) content of the article.
  • Content is run through a modern NLP javascript engine from retextjs ( and finds core elements.
  • The core themes and keywords are separated out from the current article and rendered into a simple to understand appendix section for your article

Challenges I ran into

  • There wasn't a formal API for extracting the plain text out of a confluence article. I had to build a processing function to enable parsing through the formatted confluence markdown to get the relevant content.

Accomplishments that I'm proud of

It works

What I learned

How to build a "serverless" forge app and leverage the confluence API to pull article data

What's next for Instant Wiki

The goal of the Instant Wiki would be to save any potential time needed to generate an appendix of content for an article.

  • Add more configurable options for the wiki footer content
  • Link to other sources beyond Wikipedia. Recommend additional content suggestions based on the parsed text and body from the Confluence page.
  • Parse through the user's Confluence space to find other articles that might be relevant to the current one being read.
Share this project: