Jeudi 8 octobre 2026

IBM Security Verify Access · IBM Verify Identity Access · IBM Security Verify Access Container · IBM Verify Identity Access Container

IBM Security Verify Access 10.0 through 10.0.9.2 and IBM Verify Identity Access 11.0 through 11.0.3 are vulnerable to cross-site scripting. This vulnerability allows an authenticated user to embed arbitrary JavaScript code in the Web UI thus altering the intended functionality potentially leading to credentials disclosure within a trusted session.

IBM Security Verify Access · IBM Verify Identity Access · IBM Security Verify Access Container · IBM Verify Identity Access Container

IBM Security Verify Access 10.0 through 10.0.9.2 and IBM Verify Identity Access 11.0 through 11.0.3 could allow an authenticated user to cause a denial of service using a specially crafted HTTP query due to improper allocation of system resources

IBM Security Verify Access · IBM Verify Identity Access · IBM Security Verify Access Container · IBM Verify Identity Access Container

IBM Security Verify Access 10.0 through 10.0.9.2 and IBM Verify Identity Access 11.0 through 11.0.3 could allow a remote authenticated attacker to bypass additional authentication workflow steps.

IBM Security Verify Access · IBM Verify Identity Access · IBM Security Verify Access Container · IBM Verify Identity Access Container

IBM Security Verify Access 10.0 through 10.0.9.2 and IBM Verify Identity Access 11.0 through 11.0.3 could allow a remote authenticated attacker to bypass security restrictions due to improper authentication.

IBM Security Verify Access · IBM Verify Identity Access · IBM Security Verify Access Container · IBM Verify Identity Access Container

IBM Security Verify Access 10.0 through 10.0.9.2 and IBM Verify Identity Access 11.0 through 11.0.3 is vulnerable to cross-site scripting. This vulnerability allows an unauthenticated user to embed arbitrary JavaScript code in the Web UI thus altering the intended functionality potentially leading to credentials disclosure within a trusted session

IBM Security Verify Access · IBM Verify Identity Access · IBM Security Verify Access Container · IBM Verify Identity Access Container

IBM Security Verify Access 10.0 through 10.0.9.2 and IBM Verify Identity Access 11.0 through 11.0.3 could allow a remote authenticated attacker to modify data and cause a denial of service due to an SQL injection vulnerability.

IBM Security Verify Access · IBM Verify Identity Access · IBM Security Verify Access Container · IBM Verify Identity Access Container

IBM Security Verify Access 10.0 through 10.0.9.2 and IBM Verify Identity Access 11.0 through 11.0.3 could allow a denial of service due to a heap-based out-of-bounds read. A remote attacker could send a specially crafted request that causes a limited out‑of‑bounds memory read.

IBM Security Verify Access · IBM Verify Identity Access · IBM Security Verify Access Container · IBM Verify Identity Access Container

IBM Security Verify Access 10.0 through 10.0.9.2 and IBM Verify Identity Access 11.0 through 11.0.3 may be vulnerable to audit log forgery.

IBM Security Verify Access · IBM Verify Identity Access · IBM Security Verify Access Container · IBM Verify Identity Access Container

IBM Security Verify Access 10.0 through 10.0.9.2 and IBM Verify Identity Access 11.0 through 11.0.3 local management interface in certain configurations is vulnerable to cross-site scripting. This vulnerability allows an authenticated user to embed arbitrary JavaScript code in the Web UI thus altering the intended functionality potentially leading to credentials disclosure within a trusted session.

MIT krb5

MIT Kerberos 5 (krb5) through 1.22.2 contains a NULL pointer dereference in make_cred_list() in rd_cred.c that allows authenticated Kerberos clients to crash services by sending mismatched KRB-CRED arrays. Attackers can send forwarded credentials with more tickets than ticket_info entries through gss_accept_sec_context() to crash GSS-API acceptor services, causing denial of service.

MIT krb5

MIT krb5 through 1.22.2 contains a NULL pointer dereference vulnerability in the KDC's get_pac_princ_with_realm() that returns success while leaving the client principal NULL on malformed names. A malicious or compromised cross-realm trusted KDC can send an S4U2Proxy request with a PAC carrying a malformed client name to crash krb5kdc and deny authentication.

indico

Indico is an event management system that uses Flask-Multipass, a multi-backend authentication system for Flask. Prior to 3.3.13, users who can create content, including speakers who can create minutes, can store crafted HTML in event minutes. When concurrent edits are made to the same minutes, the minute editor conflict UI can execute attacker-controlled script in the viewer's browser in the Indico origin. This issue is fixed in version 3.3.13.

MongoDB Python Driver

The MongoDB Python Driver's binary accelerator can read outside a buffer when an application decodes malformed BSON containing a truncated regular-expression element without a trailing NUL byte. An actor who can supply BSON to the documented decode or decode_all API can cause the application process to terminate when the C extension is loaded. The driver's normal database wire-protocol path does not reach this code.

TP-Link System Inc. Tapo S505 v1.6

Affected Tapo device firmware stores device-specific cryptographic material in plaintext within nonvolatile storage. An attacker with physical access to an affected device can recover this sensitive material from the firmware.  Successful exploitation of this vulnerability may result in the disclosure of device-specific cryptographic material and could, under certain conditions, increase the risk of unauthorized access to related protected information or communications.

IBM Guardium Data Protection

IBM Guardium Data Protection 12.2.2 is affected by a sensitive information exposure vulnerability. During SECRET and API_KEY rotation processing, sensitive credential material is logged at INFO level by the edge-controller/edge-manager components. An authenticated attacker with access to the relevant application or container logs could obtain these credentials and use them to impersonate services or gain unauthorized access to the Guardium control plane.

IBM Guardium Data Protection

IBM Guardium Data Protection 12.2 IBM Security Guardium Data Protection is vulnerable to stored cross-site scripting (XSS) in the Quick Search results grid. An unauthenticated attacker who can influence monitored database traffic could execute malicious script in the browser of an authenticated Guardium user.

dolibarr

Dolibarr ERP CRM before 24.0.2 contains an incorrect authorization vulnerability in htdocs/core/ajax/updateextrafield.php that checks only read permission before writing extrafield values. Authenticated users with read-only access can POST objectType, objectId, field and value parameters to persistently modify extrafields on viewable third parties, products, members, projects or contacts.

freedesktop poppler

Poppler 0.42.0 through 26.10.0 contains a stack-based buffer overflow in Decrypt::revision6Hash() that allows attackers controlling the password to overwrite stack memory when opening AESV3/R6 encrypted PDFs. Attackers can supply a password longer than 127 bytes through applications using the libpoppler, libpoppler-glib or C++ API to overflow the K1 and E buffers, crashing the process or corrupting memory.

aws-cdk-lib

Improper link resolution before file access in the asset bundling output handling in AWS aws-cdk-lib before 2.267.0 might allow a context-dependent actor to cause files from the build host to be published as the deployed asset. To remediate this issue, users should upgrade to version 2.267.0 or later.

indico

Indico is an event management system that uses Flask-Multipass, a multi-backend authentication system for Flask. Prior to 3.3.13, users who can manage events or create content, including speakers who can upload material, can store crafted javascript URLs in fields that accept custom URLs. A user who follows one of these URLs can execute attacker-controlled script in the user's browser in the Indico origin. This issue is fixed in version 3.3.13.

indico

Indico is an event management system that uses Flask-Multipass, a multi-backend authentication system for Flask. Prior to 3.3.13, an authenticated user can misuse the legacy session export API to retrieve details for a restricted session without access to that session, as long as the containing event is accessible. The missing access check can disclose session metadata such as the title, description, and conveners. This issue is fixed in version 3.3.13.

indico

Indico is an event management system that uses Flask-Multipass, a multi-backend authentication system for Flask. Prior to 3.3.13, the previous fix for CVE-2026-25738 did not cover an edge case, allowing an event organizer to submit a crafted URL that points to a prohibited local target but is accepted as valid by Indico. The organizer can read data returned by the target through affected Indico features. This issue is fixed in version 3.3.13.

freescout-help-desk freescout

FreeScout is a self-hosted help desk and shared mailbox. Prior to 1.8.235, when APP_CLOUDFLARE_IS_USED is enabled, FreeScout trusts an unvalidated CF-Connecting-IP header during failed login attempts and stores the spoofed value in the activity log. LogsMonitor inserts the value into an administrator alert email without HTML escaping, allowing injected HTML to execute when an administrator opens the email. This issue is fixed in version 1.8.235.

Borewit music-metadata

music-metadata is a metadata parser for audio and video media files. Prior to 11.16.0, the Matroska and WebM EBML parser decodes an attacker-controlled VINT element length and uses it for string-token or Uint8Array allocation before confirming that the leaf fits within its parent or available input. Crafted WebM, MKV, or MKA inputs can cause disproportionate allocations, out-of-memory denial of service, or, for a demonstrated parseFile path on Node.js 26.7.0, an uncatchable V8 fatal abort. The exact failure mode depends on the tokenizer, parser API, and runtime, but the affected leaf-length validation flaw is shared and has availability impact only. This issue is fixed in version 11.16.0.

MongoDB PHP Driver

The BSON encoder in the MongoDB PHP Driver does not check some return values after a document exceeds libbson's size limit. This can leave the encoder in an invalid state. An unauthenticated actor who can cause an affected application to encode an unusually large data structure can terminate the PHP worker or cause the resulting document to omit fields. No MongoDB server connection or database authentication is required.

An issue in gnutls v3.8.13 causes legitimate CA certificates to be rejected, leading to a Denial of Service (DoS).

IBM Guardium Data Protection

IBM Guardium Data Protection 12.0, 12.1, 12.2 is affected by an improper validation of user-supplied pointers in the WfpMonitor kernel driver. Certain METHOD_NEITHER IOCTL handlers dereference user-controlled pointers without adequate probing and exception handling, potentially allowing a privileged local attacker to cause a system crash or perform limited kernel-memory reads.

Borewit music-metadata

music-metadata is a metadata parser for audio and video media files. Prior to 11.16.0, the ID3v2 parser trusts the syncsafe tag-size field and allocates the complete tag body before checking whether the input contains the declared bytes. A truncated file containing only an ID3v2 header can request an allocation approaching 268 MiB; the allocation succeeds, the subsequent read reaches end of stream, the EndOfStreamError is caught internally, and the caller receives a normal metadata object. This issue is fixed in version 11.16.0.

Borewit music-metadata

music-metadata is a metadata parser for audio and video media files. Prior to 11.16.0, the APEv2 parser reads an attacker-controlled tag-item size and allocates a Uint8Array for a binary item before proving that the declared item fits in the remaining tag or file data. A small crafted APE file can therefore trigger a disproportionate allocation, including through cover-art items, and repeated or concurrent parsing can exhaust process memory. The demonstrated impact is availability loss only. This issue is fixed in version 11.16.0.

rabbitmq amqp091-go

amqp091-go is a Go AMQP 0.9.1 client. From 1.13.0 until 1.14.0, the frame-size mitigation from the prior allocation advisory can be bypassed before connection.tune completes because Connection.maxFrameSize uses zero for both the not-yet-negotiated and negotiated-unlimited states. A malicious or compromised AMQP peer can send a short body-frame header with a large declared payload length, causing ReadFrame and the body-frame parser to allocate attacker-selected memory before the payload is received or the frame's protocol state is rejected. The condition is reachable through public Open even when Config.FrameSize is set to the protocol minimum and can cause severe memory pressure, out-of-memory termination, or loss of the client process before authentication completes. This issue is fixed in version 1.14.0.

mariadb-corporation mariadb-connector-nodejs

MariaDB Connector/Node.js is used to connect applications developed on Node.js to MariaDB and MySQL databases. From 3.3.0 until 3.5.4, the zero-configuration TLS fingerprint-validation path calls Ed25519PasswordAuth.hash() through Authentication.validateFingerPrint, but Ed25519PasswordAuth.hash() references a seed identifier that is not in scope. Exposure requires a MariaDB server reached over TCP, TLS enabled with ssl: true or an ssl object whose rejectUnauthorized value is not false, a password set, no ssl.ca configured, and client_ed25519 negotiated as the authentication plugin. Under those conditions, a legitimate server, malicious server, or network attacker presenting a self-signed certificate can reach this path and cause a synchronous ReferenceError to escape the socket data handler. Under Node.js default uncaught-exception behavior, the client process terminates, causing denial of service. Configurations using a provided CA, rejectUnauthorized: false, another authentication plugin, or a Unix socket do not reach this vulnerable path. This issue is fixed in version 3.5.4.

MongoDB Go Driver

Improper validation of a BSON array length in the MongoDB Go Driver can cause an out-of-bounds index and runtime panic when an application calls bson.RawArray.Validate or bsoncore.Array.Validate on a malformed four-byte array. An unauthenticated actor who can supply raw BSON array data to an affected application may terminate an unprotected application process, causing a denial of service. No confidentiality or integrity impact has been identified.

MongoDB Go Driver

An integer overflow in BSON value-length handling in the MongoDB Go Driver can cause a runtime panic when an application validates or accesses a malformed BSON document. An unauthenticated actor who can supply BSON bytes to an affected application may terminate an unprotected application process, causing a denial of service. The driver's server-monitoring path contains panic recovery and is limited to server-selection failure.

MongoDB C Driver

An incorrect calculation in Decimal128 string parsing in the MongoDB C Driver can accept certain over-precision inputs containing leading zeros instead of rejecting them. This produces a value different from the supplied text. An actor who can provide a decimal string to an embedding application, including through Extended JSON parsing, can cause the application to store or use an incorrect numeric value.

MongoDB C Driver

The BSON buffer-reservation API in the MongoDB C Driver can record a length smaller than the five-byte BSON minimum. Later append or comparison operations can underflow unsigned length calculations and read or write outside the document buffer. An actor who can influence the length supplied by an embedding application can cause the application to terminate or read or corrupt adjacent process memory. Reaching this issue requires the application to pass an undersized value to bson_reserve_buffer and then perform an affected operation.

MongoDB libmongocrypt

The explicit decryption component of MongoDB libmongocrypt can return an unrecognized encrypted payload unchanged instead of returning a decryption error. An actor who can modify stored encrypted fields, such as a database writer, server, or network intermediary, can cause an affected application to process the supplied bytes as decrypted plaintext.

MongoDB C Driver

An off-by-one error in the BSON bulk document writer in the MongoDB C Driver can write one zero byte immediately past a heap allocation when a document ends at a specific buffer boundary. An actor who can influence the size of documents serialized by an embedding application can corrupt adjacent process memory or terminate the process. Reaching this issue requires the application to use the BSON bulk-writer API and produce a precise cumulative document size.

MongoDB C++ Driver

The MongoDB C++ Driver discards content after an embedded NUL byte in certain field and collection names accepted by the collection API. This can cause the driver and the calling application to interpret the same name differently. An authenticated actor who can influence a name passed by an affected application can cause the application to read distinct values from an unintended field or rename an unintended collection. These operations use the application's existing database credentials.

MongoDB libmongocrypt

An integer underflow in the KMS endpoint-parsing logic of MongoDB libmongocrypt can cause an allocation failure that terminates the application process. This can occur when an authenticated user modifies a key document in the key vault collection, or when an application accepts a KMS endpoint containing a colon after its path or query during key creation. The issue does not access memory outside its allocated bounds.

Docker Sandboxes

Docker Sandboxes compared OAuth token-endpoint hostnames case-sensitively when deciding whether to mask managed credential responses, while request routing treated DNS hostnames case-insensitively. Untrusted code inside a sandbox could use a case-variant hostname to reach the genuine provider endpoint while bypassing response masking. If a user completed the OAuth flow, the provider's access and refresh tokens could be returned unmasked to the sandbox, exposing host-managed credentials.

Docker Sandboxes

Docker Sandboxes could forward a client-supplied credential alongside a credential injected by the host egress proxy. The proxy removed alternate credentials only when their values matched known sentinel values, so untrusted code in an authorized sandbox could supply an unrecognized credential in another supported authentication header. For affected upstream services, this could authenticate the request to an attacker-controlled account and expose data included in the request.

Docker Sandboxes

Docker Sandboxes could fail open while masking credentials in protected proxy responses. When a response-body read returned data together with an error, affected handlers could forward unmasked bytes. Code inside an authorized sandbox could use this to recover host-managed OAuth access and refresh tokens or a derived Anthropic API key intended to remain outside the sandbox.

Improper certificate validation in gnutls v3.8.13 causes the application to accept certificates containing invalid extensions.

Improper certificate validation in gnutls v3.8.13 causes the application to reject legitimate certificates for valid users, leading to a Denial of Service (DoS).

OpenStack Mistral

In OpenStack Mistral through 23.0.0, the workflow membership API lets a project that has accepted a share of another project's private workflow create a further membership naming a third project. The new membership row is created with its project_id defaulted to the accepting project rather than the original workflow owner, and thus the owner can neither see nor delete it. The third project can accept this membership (that it had not actually been granted by the owner), and then read and execute the owner's private workflow; only the accepting (not the owning) project can later revoke that access.

Webkul QloApps

QloApps through 1.7.0 contains an authorization bypass vulnerability in AdminHotelRoomsBookingController::postProcess() that allows restricted back-office employees to access other hotels' data by supplying an id_hotel parameter. Attackers can modify the id_hotel URL parameter on the Book Now page to view room availability and booking status of hotels outside their assigned profile access.

FFmpeg · Ffmpeg ffmpeg

FFmpeg before 7.1.4 and 8.0.x before 8.0.2 contains a server-side request forgery vulnerability in ff_rtsp_connect() in libavformat/rtsp.c that follows RTSP 3xx redirects without validating the Location URL. Malicious RTSP servers can redirect FFmpeg to internal hosts and ports under other schemes, bypassing -protocol_whitelist, to probe internal network services.

FFmpeg · Ffmpeg ffmpeg

FFmpeg before 8.1.3 contains a protection mechanism failure in the HLS demuxer that allows attackers to bypass protocol and allowed_extensions restrictions when opening child playlists. Attackers can supply a crafted master playlist whose child playlists use disallowed protocols or non-multimedia local files, making parse_playlist() open resources the HLS security policy should block.

FFmpeg · Ffmpeg ffmpeg

FFmpeg through 9.0.2 contains an infinite loop vulnerability in ff_rtsp_connect() in libavformat/rtsp.c that follows RTSP 3xx redirects without any redirect limit. Attackers controlling an RTSP server can answer every request with a 302 redirect to itself or another server, causing endless reconnects that saturate a CPU core.

FFmpeg · Ffmpeg ffmpeg

FFmpeg before 8.1.3 contains an infinite loop vulnerability in the HLS demuxer that allows remote attackers to cause denial of service because parse_playlist() accepts Master Playlist tags inside Media Playlists. Attackers can trick victims into opening a crafted self-referencing playlist that endlessly adds variants in hls_read_header(), causing unbounded CPU and I/O consumption.

darylldoyle svg-sanitizer

savg-sanitizer is a PHP SVG/XML sanitizer. Prior to 1.0.0, svg-sanitizer's isHrefSafeValue() validates an SVG href after XML DTD entity expansion, but saveXML() serializes the original entity reference after removing the DTD declaration. A crafted entity such as Tab can appear to the sanitizer as a safe fragment prefix while HTML5 Named Character Reference resolution during inline HTML rendering later converts the surviving reference to whitespace, exposing a javascript: URL. When an application embeds the sanitized SVG inline, a user who activates the link can cause script to execute in the embedding page's origin. This issue is fixed in version 1.0.0.

darylldoyle svg-sanitizer

savg-sanitizer is a PHP SVG/XML sanitizer. Prior to 1.0.0, svg-sanitizer allows a crafted SVG DTD with a #FIXED attribute default to make cleanAttributesOnWhitelist() perform a double DOMElement::removeAttribute() call on the same attribute name in src/Sanitizer.php. The first removal deletes the explicit attribute, while the DTD default rematerializes the value before the href safety path performs the second removal, which can corrupt libxml state and terminate the PHP worker. An attacker who can submit SVG content to a sanitization endpoint can repeatedly interrupt workers and degrade or exhaust application availability. This issue is fixed in version 1.0.0.

CISA Malcolm

The Arkime live capture service (arkime-live) in Malcolm runs with network_mode: host, exposing port 8005 on all network interfaces (viewHost=0.0.0.0). Arkime trusts the X-Forwarded-User header from any IP address (userAuthIps=::,0.0.0.0/0) and auto-creates users with full access. The passwordSecret is hardcoded to the public value "Malcolm". A network-adjacent attacker bypasses nginx entirely by connecting directly to port 8005 with a forged identity header.

CISA Malcolm

Malcolm's front nginx reverse proxy defines a "Dashboards → Arkime shortcut" location using a case-insensitive regex matcher but a case-sensitive rewrite. A request whose path segment is not exact-lowercase (for example /IDDASH2ARK/...) enters the location (the matcher fires) but evades the rewrite (no redirect is issued), so nginx falls through to the location's proxy_pass to the Arkime backend. That location is the one proxied location in the shipped config that does not include the per-location authentication file, so the request reaches Arkime unauthenticated. The same location also forwards a client-supplied X-Forwarded-User header un-overwritten, and Arkime is configured to trust X-Forwarded-User as the authenticated username — so an unauthenticated network caller can reach the Arkime backend while supplying a forged, auto-provisioned identity.

CISA Malcolm

Malcolm's upload-processing pipeline (scripts/safe-extract.py) enforces entry-count, nesting-depth, and total-uncompressed-byte limits when extracting container archives (zip/tar/rar/7z via libarchive), but those limits are not applied when the uploaded file is a single-stream compressed format (.gz, .bz2, .xz, .lzma, .lz) that isn't a .tar.*-style archive. Any authenticated user permitted to upload PCAP/log files can upload a small, highly compressible file (e.g. a gzip bomb) that decompresses to an effectively unbounded size on disk, exhausting the shared Docker volume used by OpenSearch, Logstash, Arkime, and Zeek, and disrupting the platform for all users.

CISA Malcolm

Malcolm's nginx Lua role-based access control (RBAC) layer decides whether an authenticated user may reach a role-restricted path (e.g. /htadmin, /auth, /admin_login, /arkime/api/esadmin, NetBox, upload endpoints) by pattern-matching the raw, percent-encoded request URI. Nginx itself, however, selects which location block actually serves the request using the percent-decoded, normalized URI. Because the RBAC check never percent-decodes its input, an authenticated low-privilege user can request an admin-only path using percent-encoding (e.g. /%68tadmin.php) and have nginx route it to the restricted location while the Lua RBAC gate evaluating the un-decoded raw string finds no matching restriction and grants access.

aws-toolkit-vscode

Insecure file permissions in the CodeCatalyst connection handler in AWS Toolkit for VS Code before 4.10.0 allowed local users to obtain CodeCatalyst bearer tokens via reading world-readable token cache files. To mitigate this issue, users should upgrade to version 4.10.0 or later.

A hostname verification bypass in GnuTLS v3.8.13 allows attackers to circumvent the Common Name fallback mechanism and eavesdrop on communications via a crafted certificate.

moby sys

The `github.com/moby/sys/user` package provides Go utilities for parsing and looking up entries in Unix-style user and group database files. Versions before 0.4.1 do not sufficiently limit entries when parsing `/etc/passwd`- or `/etc/group`-style files, allowing an attacker who can supply a specially crafted file to cause excessive memory consumption and potentially terminate the affected process due to an out-of-memory condition. This issue is patched in version 0.4.1. As a workaround, avoid parsing attacker-controlled user or group database files, or validate and limit untrusted input before parsing it.

Zimbra Collaboration Suite

A policy-enforcement flaw in Zimbra Collaboration Suite allows an authenticated user to bypass disabled mail forwarding by using a Sieve notify action to send copies of email content and headers to an arbitrary address.

mcollina msgpack5

msgpack5 is a msgpack v5 implementation for node.js and the browser. Prior to 6.1.0, constructing msgpack5 with an empty or partial options object disables the default protoAction: 'error' protection. A decoded map containing a __proto__ key can then replace the decoded object's prototype, potentially changing inherited properties or downstream behavior, although Object.prototype is not modified globally. This issue is fixed in version 6.1.0.

Zimbra Collaboration Suite

An authorization bypass in the EWS FindItem handler of Zimbra Collaboration Suite 10.1.0 through 10.1.19 allows an authenticated user with EWS enabled to read complete mailbox items, including raw MIME and attachments, from arbitrary local accounts without a share or delegation grant.

mcollina msgpack5

msgpack5 is a msgpack v5 implementation for node.js and the browser. Prior to 6.1.0, the streaming decoder treats the reserved MessagePack byte 0xc1 as incomplete input instead of invalid input. When 0xc1 begins a stream, subsequent data remains buffered while the decoder waits for bytes that cannot make the value valid, allowing a remote peer to exhaust memory. This issue is fixed in version 6.1.0.

mcollina msgpack5

msgpack5 is a msgpack v5 implementation for node.js and the browser. Prior to 6.1.0, the array and map decoding paths have no nesting-depth limit, allowing an attacker who can provide MessagePack input to submit deeply nested containers that exhaust the JavaScript call stack and interrupt a process, worker, or request handler. This issue is fixed in version 6.1.0.

mcollina msgpack5

msgpack5 is a msgpack v5 implementation for node.js and the browser. Prior to 6.1.0, the streaming decoder reparses an incomplete array or map from the beginning whenever another chunk arrives. A remote peer can split one valid MessagePack container across many small chunks, causing completed elements to be decoded repeatedly, producing quadratic CPU use and blocking the event loop. This issue is fixed in version 6.1.0.

pydantic-ai · pydantic-ai-slim · Pydantic pydantic ai

Pydantic AI is a Python agent framework for building applications and workflows with Generative AI. From 1.77.0 until 1.107.2 and 2.24.0, web_fetch_tool, the WebFetch local fallback, and remote FileUrl media downloads buffer the complete HTTP response body before enforcing content-size controls. An attacker-influenced URL can stream an arbitrarily large response that exhausts process memory and crashes the worker; affected media types include ImageUrl, DocumentUrl, VideoUrl, and AudioUrl. SSRF protections remain effective, and the impact is limited to availability. This issue is fixed in versions 1.107.2 and 2.24.0.

pydantic-ai · pydantic-ai-slim · Pydantic pydantic ai

Pydantic AI is a Python agent framework for building applications and workflows with Generative AI. From 0.3.4 until 1.107.4 and 2.27.1, OpenTelemetry instrumentation configured with InstrumentationSettings(include_content=False) can export retry prompts outside tool calls in gen_ai.input.messages and pydantic_ai.all_messages. Agents using NativeOutput, PromptedOutput, or output validators on text output can therefore disclose validation feedback, including invalid model values quoted by that feedback, to readers of the telemetry backend. Tool-call retries and deployments that do not use include_content=False are not affected by this specific path. This issue is fixed in versions 1.107.4 and 2.27.1.

pydantic-ai · pydantic-ai-slim · Pydantic pydantic ai

Pydantic AI is a Python agent framework for building applications and workflows with Generative AI. From 1.34.0 until 2.30.0, the Agent.to_web() and clai web development chat server does not validate the Host header, allowing a website visited by a developer to use DNS rebinding to reach a loopback-hosted agent as a same-origin service. The hostile page can read the served UI and submit chat requests that execute agent tools with the local process's privileges and credentials, causing data disclosure or unwanted side effects. Binding to localhost, Origin checks, and CSRF tokens do not prevent the same-origin DNS rebinding path. This issue is fixed in versions 1.107.5 and 2.30.0.

pydantic-ai · pydantic-ai-slim · Pydantic pydantic ai

Pydantic AI is a Python agent framework for building applications and workflows with Generative AI. From 0.3.4 until 1.107.6 and 2.44.0, OpenTelemetry instrumentation configured with InstrumentationSettings(include_content=False) can still export sensitive agent content through exception.message and exception.stacktrace events, error status descriptions, and model_request_parameters containing instructions or the prompted_output_template. The exposed data is available to readers of the configured telemetry backend and can include tool feedback, provider error bodies, runtime instructions, and structured-output templates even though message attributes are redacted. This issue does not grant new access to agent data, and deployments that do not use include_content=False are not affected by the setting bypass. This issue is fixed in versions 1.107.6 and 2.44.0.

pydantic-ai · pydantic-ai-slim · Pydantic pydantic ai

Pydantic AI is a Python agent framework for building applications and workflows with Generative AI. From 1.77.0 until 1.107.6 and 2.44.0, the local web_fetch_tool and the WebFetch local fallback process server-controlled responses with quadratic title extraction, whitespace normalization, and ordered-list numbering. An attacker-controlled page of modest size can therefore block the event loop for an extended period, stalling other agent runs and requests, while unsupported codecs or excessive HTML or JSON nesting can abort an individual run. This issue is fixed in versions 1.107.6 and 2.44.0.