GHSA-pq4g-8rxf-4g29 CVE-2026-116695
Crass: Incorrect tokenization of bad URLs enables CSS sanitizer bypass
- Severity
- 0.0 / 10
- Medium · CVSS 4.0
- Exploit chance (EPSS)
- 0.35%
- Chance of exploitation in the next 30 days. Higher than 29.04% of scored vulnerabilities.
- Known exploited
- No
- Not in CISA's catalog
- Weakness
- CWE-436 Interpretation Conflict
- Fix
1.0.8- Upgrade crass to this version or later
Description
Summary
Crass's tokenizer didn't terminate a bad URL token at a ) that followed an escaped backslash (\\). As a result, Crass consumed more input as part of the bad URL than a spec-compliant browser does. CSS sanitizers that parse untrusted CSS with Crass and re-emit it could be tricked into passing through declarations, style rules, or at-rules (including @import) that the sanitizer would otherwise have removed, but that a browser would then parse and apply.
Details
While consuming the remnants of a bad URL (CSS Syntax Level 3, §4.3.15), Crass checked whether the previously consumed code point and the next code point formed a valid escape, in addition to checking the next two unconsumed code points. After an escaped backslash, the previously consumed code point is a backslash, so a following ) was incorrectly treated as escaped and the bad URL token continued until the next unescaped ) or the end of the input.
Browsers terminate the bad URL at that ), so everything Crass treated as part of the bad URL is live CSS to a browser.
This bug has existed in every released version of Crass.
Impact
Crass by itself only mis-tokenizes input. The security impact is on CSS sanitizers that use Crass to make allow/deny decisions and then re-serialize the CSS. Exploitation requires a sanitizer that doesn't reject bad URL tokens wherever they occur. Confirmed examples (verified in Chromium against the sanitizers' output):
- Sanitize (all versions through 7.0.0):
- In
Sanitize::CSS.propertiesandstyleattributes, a bad URL inside a[...]or(...)block in the value of an allowlisted property can smuggle arbitrary declarations, such asposition: fixedor abackground-imageURL with a disallowed protocol. - In
Sanitize::CSS.stylesheetand<style>elements, a bad URL in a selector can smuggle arbitrary style rules and at-rules, including@import.
- In
- Loofah (
Loofah::HTML5::Scrub.scrub_css, used forstyleattributes), and therefore rails-html-sanitizer (sanitize_css, andsanitizewhenstyleis a permitted attribute): a bad URL inside an allowlisted function such ascalc()orrgb()can smuggle declarations Loofah normally strips, such asposition: fixedandurl()values.
Smuggled CSS can be used for UI redress (clickjacking or phishing overlays), loading attacker-controlled external resources, and, where rules or @import can be smuggled, CSS-based data exfiltration. This vulnerability doesn't allow script execution in modern browsers.
Patches
Fixed in Crass 1.0.8. Users of Sanitize, Loofah, rails-html-sanitizer, or any other library that uses Crass to sanitize CSS should upgrade Crass to 1.0.8 or later.
Workarounds
There are no practical workarounds other than upgrading. Rejecting any CSS input that contains the character sequence \\ would block the known trigger, but this is a blunt measure and may reject legitimate CSS.
References
- CSS Syntax Module Level 3, "Consume the remnants of a bad url": https://drafts.csswg.org/css-syntax-3/#consume-remnants-of-bad-url
Affected versions
RubyGems
| Package | Vulnerable | Fixed in |
|---|---|---|
| crass | < 1.0.8 |
1.0.8 |
How to fix
Move to the fixed version or later with your package manager, update the lockfile, and run your tests. If the package only arrives through another dependency, upgrade that dependency first. If none of its releases pulls a fixed version yet, force the version with your package manager's override.
crass RubyGems → 1.0.8
- bundler
bundle update crass # Gemfile: gem "crass", ">= 1.0.8"
Severity breakdown
CVSS 4.0 0.0
CVSS:4.0/AV:N/AC:L/AT:N/PR:N/UI:P/VC:N/VI:N/VA:N/SC:L/SI:L/SA:N
- Attack vector
- Network
- Attack complexity
- Low
- Attack requirements
- None
- Privileges required
- None
- User interaction
- Passive
- Confidentiality (vulnerable system)
- None
- Integrity (vulnerable system)
- None
- Availability (vulnerable system)
- None
- Confidentiality (subsequent systems)
- Low
- Integrity (subsequent systems)
- Low
- Availability (subsequent systems)
- None
Weakness
CWE-436: Interpretation Conflict
Product A handles inputs or steps differently than Product B, which causes A to perform incorrect actions based on its perception of B's state.
References
Advisories
Packages
Other links
Timeline
- Published
- Last updated by the source
- Latest EPSS score
Related advisories
| Advisory | Severity | Why it's related |
|---|---|---|
| GHSA-8vfg-2r28-hvhj Crass: Non-ASCII characters cause superlinear CPU consumption | Medium | Also affects crass |
| GHSA-wwpr-jff3-395c Crass: A large number of adjacent CSS comments can trigger a SystemStackError | Medium | Also affects crass |
| GHSA-6jxj-px6v-747w Crass: Deeply nested CSS blocks and functions can trigger a SystemStackError or excessive memory usage | Medium | Also affects crass |
| GHSA-6wmf-3r64-vcwv Crass : Large numeric exponents cause CPU and memory denial of service | High | Also affects crass |
| GHSA-5x38-2vff-x55p Forge: X.509 basicConstraints bypass in verifyCertificateChain() via deprecated extension OIDs 2.5.29.10 and 2.5.29.13 | High | Same weakness, CWE-436 |
| GHSA-9v7c-q64q-vv33 Nuxt: Dot segments in the request path bypass `server/middleware` path checks and server route rules on page routes | Medium | Same weakness, CWE-436 |
| GHSA-m665-v7pm-7grj zip: Critical Parser Differential & Security Control Bypass via Dual-Header Inconsistencies (Zip Cloaking) Leading to Arbitrary File Overwrite & RCE | Critical | Same weakness, CWE-436 |
| EEF-CVE-2026-115700 SSRF allowlist bypass via percent-encoded bracketed IPv6 or trailing-dot IP literal in hackney | Not rated yet | Same weakness, CWE-436 |
MergeHarbor checks every repository's dependencies against this database, ranks what matters by exploitability and reachability, routes it to the owning team and opens the fix pull request.
Check your repos with MergeHarbor