CVE-2026-100676 in stoatchat
Summary
by MITRE • 09/26/2026
January, the media proxy/embed service of stoatchat (stoatchat/stoatchat), before version 0.15.5 improperly resolves SVG <image href> values as local filesystem paths when a fetched resource is served as image/svg+xml. An unauthenticated remote attacker who causes the service to proxy an attacker-hosted SVG (e.g. via the /proxy endpoint) can determine whether local files exist through observable response-time differences, and can cause supported local image files to be disclosed after re-encoding. Because each referenced file is read in full with no effective limit on the number or total volume of reads, a single request can also generate an unbounded amount of local filesystem I/O and memory pressure (the published proof of concept drives about 4.34 GB of reads), leading to denial of service. The issue is fixed in 0.15.5.
If you want to get best quality of vulnerability data, you may have to visit VulDB.
Analysis
by VulDB Data Team • 09/26/2026
The vulnerability identified in stoatchat versions prior to 0.15.5 represents a critical flaw within the media proxy and embed service, specifically affecting how SVG images with embedded external resources are processed. This component, accessible via endpoints such as /proxy, is designed to fetch remote content and serve it to clients. However, when an attacker provides an SVG file served as image/svg+xml that contains an <image href> tag pointing to a local filesystem path rather than a valid URL, the service fails to properly validate or sanitize this input. Instead of rejecting the request due to the invalid protocol scheme for a remote fetch operation, the application interprets the href value as a direct reference to the underlying operating system's file system. This behavior stems from an improper resolution mechanism that treats local paths with equal validity to web URLs when processing SVG image tags, creating a significant security gap in input validation and path traversal controls.
The operational impact of this flaw is twofold, encompassing both information disclosure and denial of service conditions. On the side of confidentiality, an unauthenticated remote attacker can leverage this misconfiguration to perform file existence checks by observing response time differences or trigger the actual reading and re-encoding of local image files for disclosure. This effectively allows the attacker to probe the server's internal filesystem structure and potentially exfiltrate sensitive data stored locally on the application server. The ability to read these files is not limited in scope, as the service attempts to process every referenced file without adequate restrictions on which directories or file types are accessible, thereby expanding the blast radius of a single malicious request into broader system compromise scenarios involving local resource exposure.
Furthermore, this vulnerability introduces severe availability risks through unbounded resource consumption. Because each referenced file is read in its entirety and there is no effective limit imposed on the number of reads or the total volume of data processed per request, an attacker can construct payloads that trigger massive I/O operations and memory allocation. Proof-of-concept demonstrations have shown that a single crafted request can drive approximately 4.34 GB of local filesystem reads, leading to significant performance degradation, high CPU usage, and potential application crashes. This constitutes a classic denial of service vector where the attacker exploits the server's willingness to process large amounts of data without rate limiting or size constraints, effectively exhausting system resources and rendering the service unavailable to legitimate users.
From a technical classification perspective, this vulnerability aligns with CWE-20 Improper Input Validation regarding the failure to correctly validate external input before processing it as a file path. Additionally, the ability to read arbitrary local files maps directly to CWE-59 Improper Link Resolution Before File Access, commonly known as Path Traversal or Directory Traversal. In terms of offensive security frameworks, this behavior is consistent with ATT&CK technique T1083 File and Directory Discovery, where an adversary explores the file system to gather intelligence about the target environment. The denial of service aspect further relates to resource exhaustion attacks often categorized under availability-focused tactics in threat modeling exercises.
To mitigate this vulnerability, organizations running stoatchat versions earlier than 0.15.5 must upgrade immediately to version 0.15.5 or later, where the issue has been resolved by implementing stricter validation for SVG image href attributes. Security teams should ensure that any media proxy services enforce strict allow-listing of permitted URL schemes and protocols, explicitly blocking local file paths such as file:// or relative filesystem references from being processed as remote resources. Implementing input sanitization layers at the reverse proxy level can also provide an additional defense-in-depth measure by filtering out malformed SVG content before it reaches the application logic. Regular security audits focusing on how external data is interpreted for resource loading are essential to prevent similar path traversal and information disclosure flaws in other components of the stack.