Attack ChainAIDerived
Hypothetical attack flow derived from CVE metadata – not a confirmed exploit chain
-
AccessAccess ComfyUI upload/asset endpoint
-
DeliveryUpload SVG with embedded script
-
ExploitFile stored and served via /view
-
ExecutionVictim opens SVG inline
-
PersistBrowser executes script in ComfyUI origin
-
ImpactHijack session and call backend APIs
Vulnerability Assessment
| Weakness |
CWE
OWASP Top 10:2025
A05:2025 - Injection
ASVS 5.0
V1 - Encoding and Sanitization Output encoding, context-aware escaping
MITRE ATT&CK
2 techniques across 2 tactics: Initial Access, Execution The technique list with ATT&CK links and the kill-chain view is part of the Basic plan and up. |
| Exploitation Conditions | Exploitation requires that the attacker can upload a file (an SVG or other XML/HTML dialect) to a ComfyUI instance running a version prior to 0.28.0, and that a victim subsequently opens or views that stored file via the /view endpoint (this is the UI:R user interaction). The attack is enabled specifically because image/svg+xml and related XML content types were missing from the dangerous-MIME set, so the file is served inline rather than as a download. Limiting factors: victim interaction is mandatory (the payload does not fire automatically), the attacker needs a reachable instance that accepts uploads, and impact is greatest on shared/multi-user or internet-exposed deployments - a single-user localhost-only instance sharply reduces who can be targeted. |
| Impact | The provided CVSS 3.1 vector (AV:N/AC:L/PR:N/UI:R/S:C/C:H/I:L/A:N, base 8.2 High) reflects a network-reachable, low-complexity, unauthenticated stored XSS whose exploitation requires victim interaction (opening the crafted file) and whose scope change captures execution crossing from server-served content into the victim's browser context. EPSS was not provided and KEV status is negative, and no POC is referenced, so real-world exploitation pressure appears low-to-moderate and opportunistic rather than mass-automated. A meaningful limiting factor absent from the raw score is deployment reality: ComfyUI is frequently run as a single-user local tool on localhost, in which case the practical attack surface is small; risk rises sharply for shared, multi-user, or internet-exposed instances where one user can upload content another user views. Treat this as a genuine priority for any team running ComfyUI on a shared or exposed origin, and as lower urgency for isolated single-operator localhost use. |
| Exploit ScenarioDerived | An attacker with access to a shared or exposed ComfyUI instance uploads an SVG file containing an embedded <script> payload, which the /view endpoint serves inline as image/svg+xml. When another operator (or an admin) opens or previews that file, the script executes in the ComfyUI origin, letting the attacker steal session context, invoke backend APIs, or pivot within the interface. No public POC is referenced, and the low attack complexity is offset by the required victim interaction. |
Get an e-mail when this CVE is exploited, patched or re-scored. Free account.
Alert meAnalysisAIv1 – Jul 31, 2026
Stored cross-site scripting in ComfyUI before 0.28.0 lets an attacker upload a malicious SVG that the /view endpoint serves inline in the application origin, executing arbitrary JavaScript in the browser of any user who opens it. Because image/svg+xml and related XML content types were missing from the dangerous-content-type list, uploaded markup rendered instead of downloading; a victim viewing the file triggers script that runs with full access to the ComfyUI session. No public exploit was identified at time of analysis and it is not listed in CISA KEV, but the fix commit and released version 0.28.0 are confirmed by the vendor advisory.
RemediationAI
Upgrade to ComfyUI 0.28.0 or later, which is the vendor-released patch that centralizes dangerous-content-type handling and forces uploaded SVG/HTML/XML to download with Content-Disposition: attachment and X-Content-Type-Options: nosniff (advisory GHSA-rj8c-c4p8-3c5h; release https://github.com/Comfy-Org/ComfyUI/releases/tag/v0.28.0; commit 96e0e3585b41e1417442eaa14ec57f7b4ffcb5e0). If immediate patching is not possible, restrict who can reach and upload to the instance by binding ComfyUI to localhost or placing it behind an authenticating reverse proxy, and do not expose it to untrusted or multi-user networks; as a compensating control, front the /view and /userdata endpoints with a proxy that rewrites Content-Type to application/octet-stream and adds Content-Disposition: attachment plus X-Content-Type-Options: nosniff for user-served files. The trade-off of forcing downloads is that any legitimately inline-rendered user content will now download instead of displaying, and restricting to localhost removes remote/multi-user access entirely.
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
ComfyUI is a Python (aiohttp/web) node-based GUI and backend for running diffusion image-generation models, with file endpoints (/view, /userdata, asset download) that serve user-uploaded content. The root cause is CWE-79 (Improper Neutralization of Input During Web Page Generation): the handler maintained a hardcoded set of 'dangerous' MIME types (text/html, application/xhtml+xml, text/javascript, text/css) that omitted image/svg+xml, application/xml, text/xml and other XML dialects. SVG is an XML document that can carry inline <script> and event handlers, so when served with Content-Type: image/svg+xml and inline disposition, browsers execute it in the serving origin. The fix centralizes detection in folder_paths.is_dangerous_content_type(), which normalizes charset/casing parameters and treats any *+xml or */xml type as dangerous, then forces Content-Disposition: attachment, Content-Type override to application/octet-stream, and X-Content-Type-Options: nosniff across /view and /userdata. CPE cpe:2.3:a:comfy-org:comfyui:*:*:*:*:*:*:*:* identifies the affected product.
Know when Comfyui 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
AV:N/AC:L/PR:N/UI:R/S:C/C:H/I:L/A:N
Network-reachable unauthenticated upload with no auth by default (PR:N), victim must open the file (UI:R), and script executing across origin into the browser justifies S:C with C:H/I:L.
Primary rating from Vendor (GitHub_M).
CVSS VectorVendor: GitHub_M
Lifecycle Timeline
2DescriptionCVE.org
ComfyUI is a modular diffusion model GUI, api and backend with a graph/nodes interface. Prior to 0.28.0, the /view endpoint served uploaded SVG files inline because image/svg+xml and related XML content types were absent from the dangerous-content-type handling, allowing stored cross-site scripting in the ComfyUI origin. This issue is fixed in version 0.28.0.
Executive ViewAI
ComfyUI, an image generation platform used for AI model inference workflows, is vulnerable to stored cross-site scripting (CVE-2026-56670) that executes arbitrary JavaScript in user browsers with full access to authenticated sessions. An attacker can upload a malicious SVG file that compromises any employee who views it, potentially leading to credential theft or unauthorized access to sensitive AI models and data. Immediate upgrade to version 0.28.0 is required to eliminate this attack vector.
Apply vendor patch.
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
Cross-Site Scripting occurs when an application accepts untrusted data and sends it to a web browser without proper validation or encoding.
Arbitrary file write via path traversal in ComfyUI before 0.30.0 lets an attacker place attacker-controlled content anyw
comfyanonymous/comfyui version v0.2.4 suffers from a non-blind Server-Side Request Forgery (SSRF) vulnerability. Rated h
A stored cross-site scripting (XSS) vulnerability exists in comfyanonymous/comfyui version 0.2.2 and possibly earlier. R
Remote code execution in ComfyUI v0.23.0 (and all prior 0.x releases) lets unauthenticated attackers run arbitrary Pytho
Stored cross-site scripting in ComfyUI (Comfy-Org) before 0.28.0 lets an attacker upload HTML or SVG to the per-user use
Unauthenticated information disclosure in ComfyUI (comfy-org/comfyui) before 0.28.0 lets remote attackers read arbitrary
Path traversal arbitrary file read in ComfyUI (comfy-org) before 0.28.0 lets remote unauthenticated attackers probe arbi
Path traversal in ComfyUI up to version 0.13.0 allows authenticated remote attackers to read arbitrary files on the serv
Path traversal in ComfyUI up to version 0.13.0 allows authenticated remote attackers to read arbitrary files via manipul
Cross-site request forgery (CSRF) in ComfyUI up to version 0.13.0 allows unauthenticated remote attackers to modify appl
Cross-site scripting (XSS) in ComfyUI up to version 0.13.0 allows authenticated remote attackers to inject malicious scr
Stored cross-site scripting (XSS) in ComfyUI's userdata endpoint (getuserdata function in app/user_manager.py) allows au
Same weakness CWE-79 – Cross-site Scripting (XSS)
View allVendor StatusVendor
SUSE
Severity: Important| Product | Status |
|---|---|
| openSUSE Tumbleweed | Fixed |
EUVD-2026-51400
NVD
CVE.org
EPSS
Vulnrichment
GitHub
VulDB
Exploit-DB