CVE-2026-32626
CriticalSummary
AnythingLLM to aplikacja, która przekształca treści w kontekst używany przez LLM podczas czatowania. W wersji 1.11.1 i wcześniejszych, aplikacja zawiera podatność XSS w fazie renderowania czatu, która może prowadzić do zdalnego wykonania kodu na hoście z powodu niebezpiecznej konfiguracji Electron.
Risk Assessment
Podatność ta może umożliwić atakującym zdalne przejęcie kontroli nad systemem operacyjnym użytkownika bez jego interakcji, co stanowi poważne zagrożenie dla bezpieczeństwa organizacji.
Recommendation
Zaleca się aktualizację AnythingLLM do najnowszej wersji, aby usunąć tę podatność oraz wdrożenie dodatkowych środków zabezpieczających, takich jak stosowanie DOMPurify do sanitizacji danych wejściowych.
Related vulnerabilities
- CVE-2026-96809Critical
The EduAdmin Booking plugin (eduadmin-booking) by MultiNet Interactive AB contains an SQL Injection vulnerability (improper neutralization of special elements in SQL commands). It allows an attacker to perform blind SQL injection. This issue affects versions from n/a before 6.0.0.
- CVE-2026-96331Critical
The Ajax Search Pro plugin (ajax-search-pro) by wpdreams contains an SQL Injection vulnerability (improper neutralization of special elements in SQL commands). It allows an attacker to perform blind SQL injection. This issue affects versions from n/a through 4.29.1.
- CVE-2026-96330Critical
The tagDiv Opt-In Builder plugin (td-subscription) by tagDiv contains an SQL Injection vulnerability (improper neutralization of special elements in SQL commands). It allows an attacker to perform blind SQL injection. This issue affects versions from n/a through 1.7.6.
- CVE-2026-96328Critical
The JNews - Pay Writer plugin (up to version 12.0.1) contains an SQL Injection vulnerability due to improper neutralization of special elements in SQL commands. This allows an attacker to perform a Blind SQL Injection attack.
- CVE-2026-96327Critical
The WPLMS plugin (versions before 1.9.9.8.2) contains an SQL Injection vulnerability due to improper neutralization of special elements in SQL commands. This allows an attacker to perform a Blind SQL Injection attack.
- CVE-2026-94503Critical
The Zombify plugin (up to version 1.7.7) contains an Unrestricted Upload of File with Dangerous Type vulnerability, allowing an attacker to upload a web shell to the web server.
- CVE-2026-93947Critical
The Traveler plugin (up to version 3.2.9) contains an SQL Injection vulnerability due to improper neutralization of special elements in SQL commands. This allows an attacker to perform a Blind SQL Injection attack.
- CVE-2026-107935Critical
A path traversal vulnerability was found in gvproxy, the network forwarder provided by the gvisor-tap-vsock package. The unauthenticated /services/forwarder/expose endpoint does not validate the caller-supplied socket path, allowing an attacker to delete arbitrary files on the host system.
- CVE-2026-107909Critical
A heap-based out-of-bounds write in the ws_read_frame function (src/bolt/ws.c) and the buffer_apply_mask function (src/bolt/buffer.c) in FalkorDB before 4.20.0 allows a remote unauthenticated attacker to cause a denial of service and possibly corrupt heap memory by sending a WebSocket frame with a 64-bit extended payload length to the Bolt port. The payload length is not bounded, and the only bounds check in buffer_apply_mask is an ASSERT(), which is compiled out in release builds, so the function XORs memory beyond the end of the receive buffer with the attacker-supplied mask key. Only deployments that enable the Bolt endpoint (BOLT_PORT, disabled by default) are affected.
- CVE-2026-107908Critical
A heap-based out-of-bounds write in the BoltReadHandler function (src/bolt/bolt_api.c) in FalkorDB before 4.20.0 allows a remote unauthenticated attacker to cause a denial of service and possibly execute arbitrary code by sending a Bolt RESET message with an attacker-chosen chunk size to the Bolt port. The handler checks the size only with ASSERT(), which is compiled out in release builds, then computes a destination pointer from the wire-supplied 16-bit size and moves buffered data up to about 64 KiB backwards past the start of the read buffer. Only deployments that enable the Bolt endpoint (BOLT_PORT, disabled by default) are affected.
Original NVD description (English source)
AnythingLLM is an application that turns pieces of content into context that any LLM can use as references during chatting. In 1.11.1 and earlier, AnythingLLM Desktop contains a Streaming Phase XSS vulnerability in the chat rendering pipeline that escalates to Remote Code Execution on the host OS due to insecure Electron configuration. This works with default settings and requires no user interaction beyond normal chat usage. The custom markdown-it image renderer in frontend/src/utils/chat/markdown.js interpolates token.content directly into the alt attribute without HTML entity escaping. The PromptReply component renders this output via dangerouslySetInnerHTML without DOMPurify sanitization — unlike HistoricalMessage which correctly applies DOMPurify.sanitize().
Vulnerability data from NVD (NIST) · CISA KEV · EPSS

