Plugin detection lists duplicate and unrecognised plugins

[wpdarren]

Browser + version:

Chrome - Version 148.0.7778.168 (Official Build) (arm64)

Operating system:

MacOS 26.3.1 (a) (25D771280a)

Extension version:

0.8.3

URL where it happened (or a representative public WordPress site that reproduces):

theremotehive.com

What you expected to happen:

I expected all of the plugins activated on my site to appear and in alphabetical order.

What actually happened:

A few observations from some initial testing:

1. There are a few plugins appearing that I do not recognise:  awb, mlr and wp-abilities  These appear when logged in and out. I do have a mu plugin installatron_hide_status_test.php and a drop-in plugin
   advanced-cache.php So they could be related.

I have left a screenshot of the plugins I do have installed.

2. When I am logged in I see Yoast listed three times, and Litespeed twice, which could be due to the drop-in plugin.

3. It would be good to list the plugins by alphabetical order.

Console errors / screenshots:

Screenshots

[jakemgold]

Thanks @wpdarren for the detailed report and screenshots — they shaped how we ended up scoping this for 1.0.

The duplicates (Yoast 3×, Litespeed 2×), the unrecognized entries (`awb`, `mlr`, `wp-abilities`), and the mu-plugin / drop-in leakage are all symptoms of the same root cause: asset-path inference from the frontend is inherently lossy. Plugins that enqueue assets from multiple folders look like multiple plugins; mu-plugins and drop-ins write code paths that look like plugin slugs; and the slugs we extract are guesses, not authoritative.

These aren't bugs we can fix cleanly — they're the nature of the technique. So rather than ship a panel that confidently displays sometimes-wrong data, in #20 we moved the Site Information panel behind an explicit opt-in on the new extension options page (`chrome://extensions` → Details → Extension options), framed as experimental. Users who turn it on get exactly what shipped before; users who never opt in never see the misleading data.

Alphabetical ordering would still be a nice polish for opt-in users, and the duplicate collapse you raised could happen heuristically — both are reasonable follow-ups if the experimental panel ever graduates to a default feature. Closing this issue as the broader scope has been reframed; new issues for specific accuracy improvements within the experimental panel are welcome.