Microsoft Edge Add-ons Verified

Notebook Web Clipper – Sticky Notes & Save Articles

fa6ac3f7-34e5-5a76-978d-75217a53e0a1 | v3.4.1
65/ 100
MEDIUM risk
No change since v3.4.0
Analyst verdict
Review before use

The AI review rates the findings as likely false positive, but the risk score (65/100) still counts them.

Analysis record

Analysed
3 days ago
Version
v3.4.1
Artifact
SHA256 FF0…E51
Source
Findings (non-IoC)

Is Notebook Web Clipper – Sticky Notes & Save Articles safe?

This extension declares no special permissions or host permissions in the supplied metadata. Its listed endpoints include allrecipes.com, aktualne.cz, and strings such as action.download and buttonpane.show; the request findings come from dist/ia-summarize.bundle.js, dist/image-card.bundle.js, and dist/image-view.bundle.js, which fit article clipping, summaries, and image cards.

The three high-severity results all use the title OBFUSCATION-invisible_unicode_payload. They point to _locales/fa/messages.json, _locales/te/messages.json, and _locales/kn/messages.json, where invisible Unicode characters can be part of Farsi, Telugu, and Kannada text. If the same pattern were found inside executable code, it would deserve concern. These paths are locale files.

The scanner also flagged network APIs such as NET-FETCH-dist/ia-summarize.bundle.js-527 and NET-XMLHTTPREQUEST-dist/image-card.bundle.js-3758. Those titles identify request code, while the listed endpoints do not identify a login, banking, cookie, or payment target. The blank developer name leaves the publisher unknown, yet the file paths explain the high-severity locale results and the bundled request results without a malware signature or browser-hijacking finding.

Evidence ledger

Ranked by severity · findings with a source location link to the code viewer

224 detail rows
Showing 25 of 224 · highest severity first

Publisher Evidence

Limited evidence

Zoho Corp

Publisher identity, store signals, distribution reach, and warning signals used for context. Treat this as supporting evidence, not a clean bill of health.

Edge does not expose the same publisher verification data as IDE stores, so this score is deliberately conservative.

33
Noisy-finding weight
x1.00
Publisher domain
zoho.com
Observed
Store verification signal
Limited signal
Limited
Extension portfolio
3
Portfolio

12 evidence rows available.

Finding Categories

3
Obfuscation

AI Security Report

AI Security Review

Evidence context: threat category none; evidence quality strong.

The extension is named “Notebook Web Clipper – Sticky Notes & Save Articles” and has the description “Clear the clutter. Clip the web. Save to Notebook.” The strongest signals are the three high-severity findings titled OBFUSCATION-invisible_unicode_payload, located at _locales/fa/messages.json:0, _locales/te/messages.json:0, and _locales/kn/messages.json:0. Those paths are locale message files for Farsi, Telugu, and Kannada. Invisible Unicode markers in language resources are valid writing-system characters in several South Asian and Middle Eastern scripts, so these titles do not establish hidden executable code.

The network findings are concentrated in bundled files. NET-FETCH-dist/ia-summarize.bundle.js-527, NET-XMLHTTPREQUEST-dist/ia-summarize.bundle.js-3758, and the related NET-FETCH findings in dist/ia-summarize.bundle.js identify ordinary request APIs inside a summarization bundle. The same pattern occurs in dist/image-card.bundle.js, including NET-FETCH-dist/image-card.bundle.js-215 and NET-XMLHTTPREQUEST-dist/image-card.bundle.js-3758, and in dist/image-view.bundle.js:215. Fetch and XMLHttpRequest calls fit an extension that clips articles, creates image cards, and summarizes saved content. The available endpoint list includes malformed-looking strings such as action.download, buttonpane.show, and classes.today, alongside recognizable sites such as allrecipes.com and aktualne.cz. Those strings lack the form of a clear command-and-control or credential-targeting domain, and the supplied network findings do not identify data sent to a login, banking, or payment service.

The scanner also reports no malware-signature match, no tool-poisoning match, and no secret finding. The absence of a developer name is a real trust gap, yet _locales/fa/messages.json:0, _locales/te/messages.json:0, and _locales/kn/messages.json:0 provide a direct explanation for every high-severity result. The dist/ paths provide a direct explanation for the repeated network detections. The extension declares no permissions or host permissions in the supplied metadata, which limits the browser access shown here.

A skeptic could point to the blank developer field, the unknown endpoint strings, and the high severity assigned to OBFUSCATION-invisible_unicode_payload. Those points justify keeping the publisher identity unresolved. They do not outweigh the file context: the obfuscation titles occur only in script-specific locale files, while the request titles occur in feature bundles named ia-summarize, image-card, and image-view. No finding title names cookie access, history manipulation, search replacement, credential collection, proxy routing, or malware delivery. A clean runtime trace or publisher verification would add assurance, but the supplied static evidence supports scanner noise rather than intentional harm.

Key Reasons

  • OBFUSCATION-invisible_unicode_payload occurs only in locale files for Farsi, Telugu, and Kannada.
  • Network findings are located in feature bundles named ia-summarize, image-card, and image-view.
  • No malware-signature, credential-theft, browser-hijacking, or tool-poisoning finding is present.
  • The listed endpoint strings do not identify a clear command-and-control or credential-targeting domain.

False Positive Considerations

  • Invisible Unicode detection in _locales/fa/messages.json, _locales/te/messages.json, and _locales/kn/messages.json.
  • Bundled network API detections in dist/ia-summarize.bundle.js, dist/image-card.bundle.js, and dist/image-view.bundle.js.
  • Malformed-looking endpoint strings such as action.download and buttonpane.show produced by IoC extraction.
  • Code-smell and IoC volume from bundled JavaScript.

Reviewed 2026-09-29; recommended action: suppress false positive; model confidence 91%.

Edge version history

Risk trend by version

2 analyzed versions. Each point is the latest successful scan for that version; failed zero-score scans are hidden. Dates are based on first seen by risky plugins.

Selected
65
Change since first
No change
Change from previous
No change
Versions:
First analyzed version
3.4.0
Aug 12, 2026
Risk range
65 to 65
Across analyzed versions
Latest analyzed version
3.4.1
Sep 28, 2026
Selected version
medium
Version
v3.4.1
3 days ago
Risk score
65
Findings
1952
Change vs previous
No change

Pick any point on the chart to explore that version's code below.

Frequently Asked Questions