Find the code
that hurts.
CodeLoupe reads your Git history inside the IDE and shows which files change most and are most complex. Those hotspots are where bugs and slowdowns collect.
Not all code is equal.
A small share of files draws most of the change, and most of the pain. In the sample repository on this page, the top 10% of files by hotspot score account for most of the churn.
How it works
Three steps, all on your machine. The first run reads the full history; reopening the project on an unchanged HEAD is instant.
Open your project
Any project in a Git repository. Multi-root projects are analyzed root by root.
~/dev/ledgerline · mainCodeLoupe reads git log locally
It parses commits in the background, follows renames and drops mass commits. Nothing leaves the machine.
… commitsSee the hotspots
A map, a sortable table and markers in the editor point to the files worth a closer look.
PaymentReconciler.kt · 0.82The whole repository on one screen
A zoomable circle-packing map. Nested circles are directories, circle size is lines of code, color is the metric you pick. Click a folder to zoom in, click a file to open it, hover for exact values.
| Path | Score | Revs | Churn | Complexity | Authors | Last changed |
|---|
More views on the same history
The map and table share one dataset. Everything below reads from that same analysis.
Project view
Hotspot files light up right in your project tree, with their revision count alongside — no separate panel to go check.
Temporal coupling
Files that keep changing together even though nothing in the code links them — often a hidden dependency across layers.
Knowledge map
Color the map by main author. A dashed outline flags files where one author holds more than 80% of the revisions — a knowledge silo.
Trends
Revisions per month as a sparkline, right in the tooltip and the table.
Export and share
Export the map as a self-contained HTML file, or copy the table as CSV — useful for a retro or a design review.
Export as HTML Copy table as CSVYour code and history never leave your machine.
CodeLoupe runs git on your computer and keeps its cache locally. It makes no network calls at all, so it works offline, behind a firewall, and on code you are not allowed to share.
Six metrics, no black box
Each number comes straight from your Git history and can be checked by hand.
Score
Normalized revisions × normalized complexity, from 0 to 1.
Revisions
Commits touching the file in the time window, excluding mass commits.
Churn
Lines added plus lines removed. Rewriting the same lines keeps adding to it.
Complexity
Measured from indentation. Deeper nesting scores higher.
Authors
Distinct committers in the time window.
Main author
The top author's share of a file's revisions. Above 80%, it's flagged as a knowledge silo.
Questions
IntelliJ IDEA and other JetBrains IDEs built on the IntelliJ Platform, version 2025.2 or newer. The project needs to be in a Git repository.
No. CodeLoupe reads git log on your machine and makes no network calls. There is no server, no account and no telemetry.
It works for every language without a parser, including config files and templates. Nesting depth tracks branching and loops closely enough to rank files, which is all a hotspot score needs.
A shallow clone is missing history, which makes the numbers misleading. Run git fetch --unshallow before analyzing. Automatic detection with an in-IDE warning is planned but not shipped yet.
See where your next refactor should start.
Install the plugin, open the CodeLoupe tool window on the right, and click Refresh.
IntelliJ IDEA and JetBrains IDEs 2025.2+ · Git repository