Skip to Content
Platform.Ontology

Spectroscopy DOM scan for extension traces

Walks the page's own DOM looking for chrome-extension:// strings that extensions inject.

Used to track youDevice detailsBrowser extensionsWeb browsersWhile you use itWhen you are signed in

In one line

Can look through its own page for traces that extensions leave behind

Walks the page's own DOM looking for chrome-extension:// strings that extensions inject.

Finds extensions that modify the page even when they hide their files from fetch probes. A passive second way to learn which add-ons the person runs while using the site.

What it can and can't do

What the user does: Nothing. It happens while the page loads and changes.

It can

  • Scan the page's own text and settings for extension web addresses
  • Spot extensions that add things to the page, such as grammar or shopping tools
  • Report any matches back to its servers

It cannot

  • See extensions that never touch the page
  • See your tabs, history or data on other sites
  • See inside an extension or its private data

How it differs between platforms: Works best in Chrome and Edge. Other browsers only partly leave these traces.

Browser support

BrowserStatusFirst versionReleasedNote
ChromeSupportedNot availableNot listedChromium only in practice
EdgeSupported792020-01-15None
FirefoxPartialNot availableNot listedWould only see moz-extension strings if scanned for
SafariPartialNot availableNot listedNone
Safari on iPhonePartialNot availableNot listedNone
Chrome on AndroidPartialNot availableNot listedNone
Samsung InternetPartialNot availableNot listedNone

How a site asks for it

APIrecursive DOM walk (scanDOMForPrefix) over text nodes and attributes; SpectroscopyEvent
How access is grantednone
What the user seesNone (no browser prompt)
Needs a secure context (HTTPS)No
Needs a user gestureNo
How long it lastsPer page; cookie spectroscopyId is session scoped
How to take it backNot listed
Inside an iframeNot listed

Typical flow

  1. Recurse through DOM nodes, text and attribute values
  2. Match the prefix chrome-extension:// and extract the 32 char ID
  3. Fire SpectroscopyEvent with the found IDs
  4. Cookie spectroscopyId set for the session (LinkedIn: used to catch malicious activity through browser extensions)

Data you get

What comes back: Extension IDs seen in DOM
Shape: list
How often: on DOM scan

FieldTypeUnitRangePrecision
extension_idslistIDs0 to a handful32 char each

Technical limits

  • Sees only extensions that inject visible traces into the DOM
  • No access to extension storage or pages

The story

What it is

A page reads its own hidden text and tags to find leftovers that your browser add-ons have injected.

For you

You never see it. Some add-ons write small marks into every page you open, like a stamp. The site reads its own page and notices the stamps.

For makers

It costs almost nothing to run and makes no extra network requests, so it is very hard to spot. Makers use it to spot bots and rule breaking tools. Critics say it also builds a profile of the person.

Where you see it

TypeExampleApps
securityA social site scans its own page for marks left by an automation add-on that scrolls feeds and copies profiles.LinkedIn
adsAn ad vendor looks for marks from ad blockers on the page to decide whether to show a polite 'please turn off your ad blocker' wall.None named
shoppingA price comparison site checks for coupon add-on marks so it can change how it shows deals.None named
analyticsA measurement team learns that 20 percent of visitors run a grammar helper that rewrites forms, and explains odd typing data.None named
securityA bank or fraud service looks for marks left by remote control or screen recording add-ons during a login.None named
gamesA game site spots add-ons that inject overlays into the page, which can be used to cheat in browser games.None named
enterpriseA company checks that staff browsers run the approved security add-on before showing a work app.None named

Risks

  • Quiet profiling
  • Hard to notice
  • Add-ons can be personal

Good uses and abuse

Legitimate uses

  • Spot extensions injecting content into the app

Known abuse

  • Profiling by installed extensions

Scores

ScoreValueMeaning
Sensitivity3 of 5How revealing the data is
Utility2 of 5How useful it is to honest apps
Friction0 of 5How much the user has to do
Fingerprint bits6Rough identifying entropy, when known

A real example: LinkedIn

Cookie table lists spectroscopyId as catching malicious activity through browser extensions. Code in the public snapshot has scanDOMForPrefix and parseFoundString for chrome-extension:// and fires SpectroscopyEvent.

Sources

How well checked: Checked against real behavior.

On this page