for your LinkedIn export
← Back to the app
About

About this experiment

Experiment and feedback

Network Atlas is an experiment, not a finished product. It has undergone only limited testing, so results may be wrong or incomplete for your data, and it is provided as is, with no warranty or support. Feedback is welcome for any future iterations. The simplest way is to open an issue on GitHub (you’ll need a free GitHub account). Please don’t include your LinkedIn data, screenshots with other people’s names, or anything private; describing what you saw is perfect.

How this works

Network Atlas is a single static web page. Your browser downloads it once (from GitHub, or you open a saved copy), and from then on everything runs on your own computer. There is no server that receives or processes your data.

How to use it

  • Drop in your LinkedIn export (the .zip, the unzipped folder, or the CSV files).
  • Your browser’s own file reader opens the files and everything is analyzed in memory on your device. Nothing is uploaded or stored.
  • Restyle it with Colors & fonts, save your summary card as an image, or save the whole report as an HTML file you can keep or share.
  • Save your report before you leave. The app stores nothing, so your results exist only while its page is open. If you close the tab, refresh it, or the app is updated behind the scenes, they are gone and you would need to drop your export in again. Save report (.html) downloads a self-contained file named like network-atlas-2026-10-04.html. It opens in any browser, works offline because it carries its own copy of the chart libraries, and keeps looking the way it did when you saved it, even if the app changes later. It is a snapshot, so it doesn’t update. It holds counts, categories and organization names, but no names of people and no message text. The AI starter kit stays in the app and is not part of the file.
  • The sample data is made up, so you can look around first.

How it is built

  • Charts use Chart.js and network graphs use D3.js. The zip is opened with JSZip.
  • The analysis is plain JavaScript. Industries are assigned by simple keyword rules, so treat them as approximate.
  • Relationship tiers come from message dates and senders only. Message text is never read.
  • The only things loaded from the internet are the chart libraries and fonts, never your data.

License

Free for noncommercial use. You may use, change and share Network Atlas, keeping the credit below, but not sell it or charge for it. Licensed under the PolyForm Noncommercial License 1.0.0. The reports you make with it are yours.

Credit

Network Atlas was built by Mary-Lynn Bragg with Claude Code. The idea and format are adapted from Visualize Your LinkedIn Data Export by Logan Currie, also built with Claude Code.