uBlock Origin's Element Picker Let Web Pages Add Trusted Filters, Fixed in 1.75.4
Security / news
uBlock Origin's Element Picker Let Web Pages Add Trusted Filters, Fixed in 1.75.4
A page could iframe the picker in the Firefox MV2 build and, after one hidden click, plant a filter that injects JavaScript. The advisory went public on November 15, twelve days after the patch.
A web page can take control of uBlock Origin's element picker in the Firefox Manifest V2 build, and after one click from the visitor it can add a trusted filter that injects JavaScript into pages. The preconditions are specific: Firefox, the MV2 extension, version 1.75.2 or lower, and at least one active filter whose scriptlet returns a web-accessible resource. The fix shipped as 1.75.4 on November 3, 2026.
The flaw is tracked as GHSA-cmqw-9r2w-6wv2, which the uBlock Origin project published on November 15 with a High rating. It carries no CVE and no CVSS score. The advisory credits a reporter listed as syvb, who sent it by email on November 1.
What an attacker needs and what they get
The attacker controls a web page the victim visits. The advisory says that if a filter uses a prevent-xhr or prevent-fetch scriptlet that returns a web-accessible resource, the page can extract the extension's UUID, the random identifier Firefox assigns to each install. With the UUID, the page can load the extension's own picker page inside an iframe.
The researcher's write-up, published November 15 by Smitty van Bodegom at iter.ca, names the filter that makes this work on a default install: lootdest.*##+js(no-fetch-if, ...) from the uAssets list. The post notes that other UUID leaks exist, so removing that one rule would not close the door.
The picker then accepts messages over a MessageChannel from whoever embedded it. In the write-up's account, the opt field of those messages can carry a newline, and the text after the newline becomes a second filter. The advisory states that filters added through the picker are treated as trusted until the next recompile. The post says trusted filters can inject JavaScript code into pages.
The victim still has to click. The post describes an iframe at zero opacity positioned so the click lands on the picker's hidden Create button. That makes this a one-click attack, not a zero-click one.
The patch timeline
The dates come from the advisory and the project's GitHub release feed.
| Date | Event | Version |
|---|---|---|
| Nov. 1 | Reported by email | n/a |
| Nov. 2 | First fix commit (bf69cf7) ships | 1.75.2 |
| Nov. 3 | Second fix commit (9859477) ships | 1.75.4 |
| Nov. 8 | Picker filters validated (b1f7d27, 150df12) | 1.76.0 |
| Nov. 15 | Advisory and write-up published | 1.76.4 is current |
The advisory lists 1.75.2 and lower as affected and 1.75.4 as patched, even though it also credits a commit in 1.75.2. The simplest reading is that the first change was incomplete and the second closed the gap. The advisory does not say so directly.
The score is missing, and the risk is narrower than High suggests
The advisory's High rating is defensible on impact, since injected script runs with a page's privileges. The preconditions pull it back. It needs Firefox, the MV2 build, a leaking scriptlet filter and a click on a hostile page. The write-up says the MV3 build, uBO Lite, has a similar but less problematic issue. It does not say whether that one is fixed.
The post also says that GPT-5.6 Sol found most of the issue, that the author verified it, and that Claude wrote most of the proof-of-concept code. That is the author's account and the advisory does not repeat it.
Neither source reports exploitation in the wild. The proof of concept is public as of November 15, and the patch had been out for twelve days by then.
What to do
Firefox users on the MV2 build should confirm they are on 1.76.4 or later. Extension updates normally arrive automatically, but the version is visible on the extension's about:addons page. After 1.76.0 the project disabled war:-related filters for the time being, so some resource-returning filters may behave differently until they are reworked.
The fix spread across three releases between November 2 and November 8. The same attack class, a web page talking to a privileged extension page, is the one documented in GitHub Security Lab's guide to attacking browser extensions.
Related reading on this site: the OpenSSH musl flaw covers another case where default configuration decided exposure, and the PyPI security tab piece covers a different way maintainers now publish provenance.
Sources
More in Security
- 01Java4Shell Scores 9.2 From Its Finder and 8.1 From OpenJDK, and Needs an AVX-512 CPUCalif says a 65-character Base64 string overwrites 62 bytes of Java heap and reaches code execution in Keycloak, Jenkins and Jira. A second security firm says the source shows only a crash.
- 02CISA Gave 9 of 16 November KEV Entries a Three-Day Deadline With Forensic Triage, and 7 Got 14 DaysThe catalog's own due-date and forensicTriage fields show two tiers under BOD 26-04. Sorting November's additions by tier shows which exploited bugs CISA treats as already-compromised-until-proven-otherwise.
- 03Project Zero Cuts Its Disclosure Deadline to 60 Days on Feb. 2, 2027, Then 30 in 2028Google's bug-hunting team is dropping the 30-day patch soak period and counting every deadline from the day it reports, citing AI-assisted exploitation.
- 04Google Finds exploit_agent Diffing Patches and Firing Exploits With No Operator in the LoopGTIG's November 16 tracker reports the first fully automated discovery-to-exploit pipeline it has seen, found on an exposed staging server, plus 22 zero-days in August.