CWE-180

A vulnerability in multiple Atlassian products allows a remote, unauthenticated attacker to bypass Servlet Filters used by first and third party apps. The impact depends on which filters are used by each app, and how the filters are used. This vulnerability can result in authentication bypass and cross-site scripting. Atlassian has released updates that fix the root cause of this vulnerability, but has not exhaustively enumerated all potential consequences of this vulnerability. Atlassian Bamboo versions are affected before 8.0.9, from 8.1.0 before 8.1.8, and from 8.2.0 before 8.2.4. Atlassian Bitbucket versions are affected before 7.6.16, from 7.7.0 before 7.17.8, from 7.18.0 before 7.19.5, from 7.20.0 before 7.20.2, from 7.21.0 before 7.21.2, and versions 8.0.0 and 8.1.0. Atlassian Confluence versions are affected before 7.4.17, from 7.5.0 before 7.13.7, from 7.14.0 before 7.14.3, from 7.15.0 before 7.15.2, from 7.16.0 before 7.16.4, from 7.17.0 before 7.17.4, and version 7.21.0. Atlassian Crowd versions are affected before 4.3.8, from 4.4.0 before 4.4.2, and version 5.0.0. Atlassian Fisheye and Crucible versions before 4.8.10 are affected. Atlassian Jira versions are affected before 8.13.22, from 8.14.0 before 8.20.10, and from 8.21.0 before 8.22.4. Atlassian Jira Service Management versions are affected before 4.13.22, from 4.14.0 before 4.20.10, and from 4.21.0 before 4.22.4.

A vulnerability in multiple Atlassian products allows a remote, unauthenticated attacker to cause additional Servlet Filters to be invoked when the application processes requests or responses. Atlassian has confirmed and fixed the only known security issue associated with this vulnerability: Cross-origin resource sharing (CORS) bypass. Sending a specially crafted HTTP request can invoke the Servlet Filter used to respond to CORS requests, resulting in a CORS bypass. An attacker that can trick a user into requesting a malicious URL can access the vulnerable application with the victim’s permissions. Atlassian Bamboo versions are affected before 8.0.9, from 8.1.0 before 8.1.8, and from 8.2.0 before 8.2.4. Atlassian Bitbucket versions are affected before 7.6.16, from 7.7.0 before 7.17.8, from 7.18.0 before 7.19.5, from 7.20.0 before 7.20.2, from 7.21.0 before 7.21.2, and versions 8.0.0 and 8.1.0. Atlassian Confluence versions are affected before 7.4.17, from 7.5.0 before 7.13.7, from 7.14.0 before 7.14.3, from 7.15.0 before 7.15.2, from 7.16.0 before 7.16.4, from 7.17.0 before 7.17.4, and version 7.21.0. Atlassian Crowd versions are affected before 4.3.8, from 4.4.0 before 4.4.2, and version 5.0.0. Atlassian Fisheye and Crucible versions before 4.8.10 are affected. Atlassian Jira versions are affected before 8.13.22, from 8.14.0 before 8.20.10, and from 8.21.0 before 8.22.4. Atlassian Jira Service Management versions are affected before 4.13.22, from 4.14.0 before 4.20.10, and from 4.21.0 before 4.22.4.

Vite is a frontend tooling framework for JavaScript. From 7.1.0 to before 7.3.2 and 8.0.5, on the Vite dev server, files that should be blocked by server.fs.deny (e.g., .env, *.crt) can be retrieved with HTTP 200 responses when query parameters such as ?raw, ?import&raw, or ?import&url&inline are appended. This vulnerability is fixed in 7.3.2 and 8.0.5.

A directory traversal vulnerability exists in Ivanti LANDesk Management Gateway through 4.2-1.9. By appending %3F.php to the URI of the /client/index.php endpoint, an attacker can bypass access controls and gain unauthorized access to various endpoints such as /client/index.php%3F.php/gsb/firewall.php within the management web panel, potentially exposing sensitive device information. NOTE: This vulnerability only affects products that are no longer supported by the maintainer.

NextAuth.js provides authentication for Next.js. Prior to @auth/core 0.41.3 and next-auth 4.24.15 and 5.0.0-beta.32, the defaultNormalizer used by the email and magic-link sign-in flow validates an address before applying Unicode normalization. An address can contain a Unicode character such as U+FF20 FULLWIDTH COMMERCIAL AT that is not ASCII at-sign but canonicalizes to an ASCII at-sign under NFKC or NFKD normalization. The address passes the normalizer's single-at-sign check, but a downstream sendVerificationRequest mail library or delivery service that normalizes the address can then see two at-sign separators and deliver the passwordless sign-in link to an attacker-controlled recipient. Applications are affected when the email provider uses the built-in normalizer rather than a custom normalizeIdentifier and the downstream sender applies Unicode normalization. An attacker who knows a victim's email address can request the misrouted magic link and sign in as the victim without victim interaction. This issue is fixed in @auth/core 0.41.3 and next-auth 4.24.15 and 5.0.0-beta.32.

Jupyter Enterprise Gateway launches remote Jupyter Notebook kernels across distributed clusters like Apache Spark, Kubernetes, and Docker Swarm. Versions 2.0.0rc1 and above prior to 3.3.0 have a prohibited UID and GID feature that by default prevents launching kernels with UID or GID 0 (root), and this restriction can be bypassed using a specially crafted KERNEL_UID or KERNEL_GID value. This input validation vulnerability allows running Jupyter kernels as root, which can be dangerous as it allows more attack surface, and may lead to container escapes, compromising the worker node and all workloads running on it. Repeated exploitation can compromise all worker nodes, and thus the entire Kubernetes cluster. It is possible to specify volume mounts, so one vector for a container escape is to use a hostPath R/W volume mount, use this UID/GID bypass to run as root, and then gain code execution in the underlying worker node by creating a crontab entry in the mounted host file system. This issue has been fixed in version 3.0.0.

The cohttp package before 6.3.0 for OCaml allows directory traversal.

jupyterlab is an extensible environment for interactive and reproducible computing, based on the Jupyter Notebook Architecture. From 4.5.0 until 4.5.10 and 4.6.2, in jupyterlab/extensions/manager.py and jupyterlab/extensions/pypi.py, JupyterLab's PyPI extension manager enforces blocked_extensions_uris by comparing requested install names to blocklist entries with custom normalization that is weaker than PyPI package-name canonicalization. An authenticated user can request a PyPI-equivalent spelling such as JupyterLab.Git for a blocklisted package such as jupyterlab-git, and JupyterLab accepts the install request even though pip resolves the variant to the same package. Security impact requires an allowlist or blocklist intended to restrict package installation, the PyPI Extension Manager, and kernels and terminals that are disabled or delegated to remote hosts. The bypass lets an authenticated user install a prohibited extension, defeat integrity restrictions, and affect availability without gaining new read access. This issue is fixed in versions 4.5.10 and 4.6.2.

Input Leap (aka input-leap) through 3.0.3, when the non-default --enable-drag-drop option is used on Windows or macOS, mishandles the / versus \ distinction and allows directory traversal, with resultant code execution if a file is written to a startup directory. This occurs via a DDRG message.

FrankenPHP is a modern application server for PHP. Prior to 1.11.2, FrankenPHP’s CGI path splitting logic improperly handles Unicode characters during case conversion. The logic computes the split index (for finding .php) on a lowercased copy of the request path but applies that byte index to the original path. Because strings.ToLower() in Go can increase the byte length of certain UTF-8 characters (e.g., Ⱥ expands when lowercased), the computed index may not align with the correct position in the original string. This results in an incorrect SCRIPT_NAME and SCRIPT_FILENAME, potentially causing FrankenPHP to execute a file other than the one intended by the URI. This vulnerability is fixed in 1.11.2.

Caddy is an extensible server platform that uses TLS by default. Prior to version 2.11.1, Caddy's FastCGI path splitting logic computes the split index on a lowercased copy of the request path and then uses that byte index to slice the original path. This is unsafe for Unicode because `strings.ToLower()` can change UTF-8 byte length for some characters. As a result, Caddy can derive an incorrect `SCRIPT_NAME`/`SCRIPT_FILENAME` and `PATH_INFO`, potentially causing a request that contains `.php` to execute a different on-disk file than intended (path confusion). In setups where an attacker can control file contents (e.g., upload features), this can lead to unintended PHP execution of non-.php files (potential RCE depending on deployment). Version 2.11.1 fixes the issue.

baserCMS is a website development framework. In versions 5.2.3 before 5.3.2 and 5.4.0 before 5.4.2, the optional BcThemeFile plugin's Admin API validates a raw path in ThemeFilesService::update() and getFullpath(), but ThemeFileForm::_execute() subsequently applies rawurldecode() to fullpath before writing. Because normalizePath() resolves only literal dot segments, percent-encoded traversal can pass the containment checks and become a path outside the theme directory at the write sink. An authenticated administrator can write PHP files into writable locations such as webroot and execute them, exceeding the intended theme-template editing boundary. The source also identifies the same check-before-decode ordering in ThemeFilesService::copy(), although that path was not dynamically verified. Exploitation requires the non-default plugin, an administrator session, and a same-origin Origin header; USE_CORE_ADMIN_API=false does not block session-authenticated requests. The Admin web form is not affected by this encoded-path tampering because FormProtection protects fullpath. This is a third bypass of the protections associated with CVE-2026-30940, using encoded traversal decoded after validation. This issue has been patched in versions 5.3.2 and 5.4.2.

`zip` is a zip library for rust which supports reading and writing of simple ZIP files. In the archive extraction routine of affected versions of the `zip` crate starting with version 1.3.0 and prior to version 2.3.0, symbolic links earlier in the archive are allowed to be used for later files in the archive without validation of the final canonicalized path, allowing maliciously crafted archives to overwrite arbitrary files in the file system when extracted. Users who extract untrusted archive files using the following high-level API method may be affected and critical files on the system may be overwritten with arbitrary file permissions, which can potentially lead to code execution. Version 2.3.0 fixes the issue.

Kestra is an open-source, event-driven orchestration platform. Prior to 1.0.45 and 1.3.23, the local internal-storage backend validates user-supplied paths for .. traversal before it converts Windows-style backslashes to forward slashes. An attacker can therefore smuggle a traversal sequence past the guard using backslashes (..\..\..\); the guard sees a harmless string, and the path is only rewritten to ../../../ after validation, immediately before the file is opened. Any authenticated user who can view an execution (the lowest-privilege role) can call GET /api/v1/{tenant}/executions/{executionId}/file?path=… and read any file on the server filesystem readable by the Kestra process, outside the storage sandbox and across every tenant and namespace. This includes the embedded H2 database (all flows, all users, all stored secrets), internal storage of every other tenant/namespace, mounted secret files, and the process environment (/proc/self/environ) which contains configured database and secret-backend credentials. It is a complete breach of Kestra's storage isolation and multi-tenancy boundary. This vulnerability is fixed in 1.0.45 and 1.3.23.

Lemonldap::NG::Handler versions from 2.0.0 before 2.16.10, from 2.17.0 before 2.21.6, from 2.22.0 before 2.23.4 for Perl allow an equivalent spelling of a path to bypass the locationRules that restrict it. The handler matches each vhost's locationRules regular expressions against REQUEST_URI, the raw request line, while the web server routes on the path it has already percent-decoded and normalized. A request that percent-encodes a character of the path, inserts dot segments, or doubles a slash therefore reaches the protected resource under a URI that no rule regexp matches, and the vhost's default rule decides access. Deny rules, identity and group conditions, and unprotect and skip rules are bypassed alike. Only a vhost whose default rule is more permissive than its other rules is affected. An authenticated user then reaches any URL a locationRules regexp was meant to restrict, but gains no more than that default rule already grants.

A vulnerability was detected in cosmicstack-labs mercury-agent up to 1.1.13. Affected is the function checkShellCommand of the file src/capabilities/permissions.ts of the component Shell Command Execution. The manipulation results in incorrect behavior order: validate before canonicalize. The attack may be launched remotely. The exploit is now public and may be used. The project was informed of the problem early through an issue report but has not responded yet.

PeaZip before 11.3.0, in a non-default configuration, is vulnerable to OS command injection via a filename in an archive because "quotation character already used in the string" is mishandled.

n8n before 2.41.4 and 2.42.x before 2.42.1 contains a signature bypass in the waiting-webhook endpoint, which selects Send-and-Wait validation before normalising an alternate node reference form. Attackers holding an execution's resume token can approve waiting executions meant for others, running downstream actions under the workflow owner's credentials, potentially including command execution.

ModSecurity is an open source, cross platform web application firewall (WAF) engine for Apache, IIS and Nginx. Prior to 3.0.16, the multipart/form-data request body parser in libmodsecurity silently removes embedded line breaks from non-file form-field values before exporting them to ARGS and ARGS_POST because src/request_body_processor/multipart.cc overwrites reserved bytes in m_reserve instead of appending the current buffer. This creates a parser differential between ModSecurity and backend applications that preserve line breaks in form fields, allowing rules that inspect ARGS or ARGS_POST to miss payloads whose dangerous syntax depends on a line break. This issue is fixed in version 3.0.16.

In Eclipse BaSyx Go Components versions up to and including 1.0.0, ABAC-enabled deployments are vulnerable to an authorization bypass caused by inconsistent trailing-slash handling between the ABAC middleware and the HTTP router. The shared router configuration used Chi's `middleware.StripSlashes`, so a request such as `GET /shells/` was dispatched to the registered `GET /shells` route. However, the ABAC middleware evaluated the original request path including the trailing slash. If ABAC route lookup did not find a matching slash-suffixed route, the request was passed onward and the router then stripped the slash and executed the protected handler without the intended ABAC authorization decision and without the expected ABAC query filters. An unauthenticated or unauthorized network attacker could append a trailing slash to protected API routes to reach handlers that should have been denied by ABAC policy. Depending on the exposed component, HTTP method, and deployed policy, this could allow unauthorized read, create, update, delete, or upload operations. The issue affects ABAC-enabled deployments of services that use the shared router and ABAC middleware, including AAS Repository, Submodel Repository, AAS Registry, Submodel Registry, Concept Description Repository, Discovery, AAS Environment upload, and related services. The issue is fixed in Eclipse BaSyx Go Components v1.0.1.

Copier is a library and CLI app for rendering project templates. From 9.5.0 through 9.16.0, percent-encoded parent-directory segments or encoded path separators in a template URL can match a configured trusted repository prefix before an HTTP server or Git transport decodes the path, allowing unsafe template features from a repository outside the trusted prefix to run after user interaction. This issue is fixed in version 9.17.0.

ProjectSend 2.0.0 before 2.7.0 contains a path validation flaw in OrphanFileScanner::isOrphan() that allows staff with import_orphans permission to treat tracked files as orphans. Attackers can submit ./ path aliases to the orphans import or delete endpoints, deleting or adopting other owners' files while bypassing delete_others_files.

OpenNext Cloudflare is an open Next.js adapter for Cloudflare. From 0.3.0 until 1.20.2, applications that enable enableCacheInterception, rely on middleware for access control, and have a matching cached protected route process encoded pathnames inconsistently between middleware matching and cache lookup. An unauthenticated attacker can use a percent-encoded pathname that does not match authorization middleware but decodes to the cache key of a protected route, causing cached content to be returned without authorization. Malformed encodings can also be partially decoded, allowing cache lookup to continue with a pathname that middleware did not evaluate consistently. This issue is fixed in version 1.20.2.

Incorrect Behavior Order: Validate Before Canonicalize in the report theme CSS sanitizer in maalfer Pentestify 1.2.0 through 2.3.2 allows an authenticated user to force outbound HTTP requests from other users' browsers, disclosing their IP address and User-Agent, via CSS hex escapes that reconstruct the url() function and evade the sanitizer blocklist

Guzzle is an extensible PHP HTTP client. Prior to 7.15.2 and 8.0.1, Guzzle gives a transport the request URI as text and supplies the Host header separately. The cURL handlers set CURLOPT_URL to the URI exactly as written and push that Host into CURLOPT_HTTPHEADER; StreamHandler does the same through fopen(). libcurl then parses the authority itself, percent-decoding it and, on an IDN-capable build, applying IDNA mapping, and uses the result to resolve, connect, name the TLS peer and address a proxy CONNECT, while the supplied Host suppresses the aligned one libcurl would have generated. For a URI host written as 127.0.0.%31, filter_var() rejects the host as an IP literal, yet libcurl decodes it to 127.0.0.1 and reaches loopback with no DNS lookup while the server receives Host: 127.0.0.%31. An attacker who influences a fetched URI can therefore reach a host the application's checks excluded and read whatever the host exposes of the response. The same divergence moves Guzzle's own decisions onto a spelling the transport does not use: no_proxy selects proxy routing from the literal host, and RedirectMiddleware decides from it whether to strip Authorization and Cookie. Exploitation requires the application to build a request URI from untrusted input and to make a host decision before handing it to Guzzle. This issue is fixed in versions 7.15.2 and 8.0.1.

@fastify/static evaluates the allowedPath callback before normalizing dot segments and duplicate path separators in the pathname used for file resolution. Versions up to and including 10.1.1 are affected. An unauthenticated attacker can bypass allowedPath restrictions by requesting equivalent non-canonical pathnames, causing files that were intended to be denied to be served anyway. The bypass does not allow access outside the configured static root by itself, it defeats path-based filtering only. The issue is patched in @fastify/static 10.1.2.

AnythingLLM is an application that turns pieces of content into context that any LLM can use as references during chatting. From 1.0.0 to 1.15.0, AnythingLLM's unauthenticated account-recovery flow in server/utils/PasswordRecovery/index.js uses recoverAccount() to deduplicate the raw recoveryCodes values before trimming them, so one valid code submitted twice with different surrounding whitespace can satisfy the two-code check. Each normalized value can also match the same stored hash instead of consuming a distinct hash. An attacker who knows the target username and one recovery code can call POST /api/system/recover-account in multi-user mode, receive a password-reset token, and use POST /api/system/reset-password to take over the account, including an administrator account.

OpenClaw versions prior to 2026.2.24 contain a path traversal vulnerability where @-prefixed absolute paths bypass workspace-only file-system boundary validation due to canonicalization mismatch. Attackers can exploit this by crafting @-prefixed paths like @/etc/passwd to read files outside the intended workspace boundary when tools.fs.workspaceOnly is enabled.

Hono is a Web application framework that provides support for any JavaScript runtime. Prior to 4.12.12, ipRestriction() does not canonicalize IPv4-mapped IPv6 client addresses (e.g. ::ffff:127.0.0.1) before applying IPv4 allow or deny rules. In environments such as Node.js dual-stack, this can cause IPv4 rules to fail to match, leading to unintended authorization behavior. This vulnerability is fixed in 4.12.12.

Vite is a frontend tooling framework for JavaScript. From 6.4.4 until 6.4.5, the Vite development server can evaluate server.fs.deny before stripping loader query parameters such as ?raw, ?import&raw, or ?import&url&inline. When the development server is exposed to the network and a sensitive file is inside server.fs.allow but matched by server.fs.deny, an unauthenticated attacker can append one of these queries and retrieve the denied file. This issue is fixed in version 6.4.5.

A hostname canonicalization vulnerability in the email HTML sanitizers of Zimbra Collaboration Suite (ZCS) allows an unauthenticated sender to bypass the restriction on forms that target the Zimbra host. DefangFilter and the OWASP ActionAttributePolicy compare the form-action hostname with the request hostname before browser URL canonicalization. A percent-encoded hostname can therefore survive sanitization while the browser submits the form to the normal Zimbra origin with the recipient's authentication cookie. This permits authenticated form requests from a crafted email after the recipient opens it in Classic and submits the form. The issue is present in ZCS 10.1.16 and 10.1.21. It does not bypass an endpoint's CSRF-token checks; a state-changing attack additionally requires an endpoint that accepts the resulting request without that protection.

SEP sesam before 5.2.0.24 mishandles User Authorization with MFA. If AD authentication is configured and MFA is enforced, an attacker can create a second OTP access capability. SEP sesam and Active Directory handle username capitalization differently, which may allow multiple SEP sesam user accounts to be created for the same Active Directory (AD) account. Active Directory treats usernames as case-insensitive, while SEP sesam distinguishes between different letter casing. As a result, the same AD user can be represented by multiple SEP sesam user accounts that differ only in username capitalization. When Active Directory authentication is configured and multi-factor authentication (MFA) is enforced, this behavior may allow an additional OTP Authenticator to be registered for the same AD account, reducing the effectiveness of MFA protection.

fast-uri is a dependency-free RFC 3986 URI parser and serializer for Node.js. In versions before 2.4.8, from 3.0.0 through 3.1.8, and from 4.0.0 through 4.2.2, fast-uri percent-decodes the scheme component with the legacy global unescape() function before validating it, so an input whose decoded scheme is a dangerous token is emitted as a live scheme even though RFC 3986 does not allow percent-encoding in a scheme and no other parser accepts it. An application that screens dangerous schemes on the raw input and then stores or renders the normalized or resolved result can turn an inert string into an executable javascript, data, or vbscript URL, producing a stored or reflected cross-site scripting vector, and the equal() comparison treats an encoded and a plain scheme as the same, which can bypass equality-based allowlists. The issue is fixed in fast-uri 2.4.8, 3.1.9, and 4.2.3, and users should upgrade to one of these versions or later. As a workaround, reject input whose scheme part contains a percent sign, or check the parsed scheme against an allowlist after normalization and use that same parsed result rather than the raw string.

A build step for a Git source, crafted in a specific way, can bypass some policy validation rules. A malicious build definition can make the repository look like it is coming from a different remote URL than it really is when Git clone is happening. If policy is doing more stricter validation, for example based on commit SHA, commit data, or signatures, then all these validations still apply correctly.

stoatchat before 0.15.5 fails to revalidate usernames after Unicode sanitization, allowing attackers to create usernames with forbidden characters by submitting Unicode letters that transform into rejected characters. Attackers can bypass character allowlists and length limits to create reserved-name lookalikes, embed special characters, and exceed the 32-character storage limit.

django-allauth before 65.19.4 does not have the expected limits on failed login attempts because, in some common configurations, an attacker can leverage the handling of diacritics (e.g., accents) for a higher effective limit.

PyJWT is a Python implementation of JSON Web Token standards. Prior to 2.14.0, PyJWT signature segment is affected because signature segment decoding accepts characters outside the canonical Base64URL representation. This occurs when non-Base64URL characters are appended to a valid compact JWS signature segment. As a result, base64url_decode produces the same signature bytes for different serialized segments. Consequently, raw-token revocation checks can fail to recognize an equivalent modified token. This issue is fixed in version 2.14.0.

Warp is an agentic development environment. From 0.2025.10.08.08.12.stable_00 until 0.2026.05.06.15.42.stable_01, Warp contains a command execution permission-check bypass in the default unsandboxed CLI agent profile. The CLI profile is non-interactive and relies on a command denylist as a safety boundary for commands that should require confirmation. Because command strings were checked before canonicalizing leading environment-variable assignments, an attacker who can influence the agent's command output may cause denylisted commands to be treated as non-denylisted. This vulnerability is fixed in 0.2026.05.06.15.42.stable_01.

Varnish Cache before 8.0.1 and Varnish Enterprise before 6.0.16r12, in certain unchecked req.url scenarios, mishandle URLs with a path of / for HTTP/1.1, potentially leading to cache poisoning or authentication bypass.

Guzzle is an extensible PHP HTTP client. Prior to 7.15.2 and 8.0.1, SetCookie::matchesDomain() gives every subdomain of a cookie Domain that cookie unless SetCookie::matchesDomain() recognizes the Domain as an IP literal or a numeric host, and the decision comes from the domain's own text, so two spellings a transport reads as an address keep subdomain scope. Hexadecimal and mixed-base forms such as 0x7f000001 and 0177.0.0.0x1 go unrecognized while libcurl 8.21.0 reads both as 127.0.0.1. A percent-escaped Domain keeps that scope on both branches because percent-decoding sits above numeric parsing, so 192.168.0.%31 and 127.0.0.1%2e are registered names in the URI grammar rather than address literals, and no numeric rule in any base classifies them, while libcurl decodes the host before resolving and reads them as 192.168.0.1 and 127.0.0.1. A cookie stored for Domain=0x7f000001 is placed in the Cookie header of a request to evil.0x7f000001, disclosing a session identifier or token to a host that is not that address, and a response from evil.0x7f000001 setting Domain=0x7f000001 is accepted into the jar and replayed to the address, so a server answering for the look-alike name can fix a session or set application state. Exploitation requires the application to enable cookie support, address an origin by one of these spellings, and contact a host whose name ends in that spelling. This issue is fixed in versions 7.15.2 and 8.0.1.

Rack is a modular Ruby web server interface. Prior to versions 2.2.23, 3.1.21, and 3.2.6, Rack::Static#applicable_rules evaluates several header_rules types against the raw URL-encoded PATH_INFO, while the underlying file-serving path is decoded before the file is served. As a result, a request for a URL-encoded variant of a static path can serve the same file without the headers that header_rules were intended to apply. In deployments that rely on Rack::Static to attach security-relevant response headers to static content, this can allow an attacker to bypass those headers by requesting an encoded form of the path. This issue has been patched in versions 2.2.23, 3.1.21, and 3.2.6.

Incorrect Behavior Order: Validate Before Canonicalize vulnerability in ash-project ash lets an attacker store a case-insensitive string value that violates its length or match constraints. Ash.Type.CiString.apply_constraints/2 (lib/ash/type/ci_string.ex) validated the max_length, min_length, and match constraints against the value as submitted, while the type case-folds the string (per its casing) for storage and comparison. Because validation ran before folding, an attacker can submit a value whose folded form breaks a constraint but whose original form passes: for example, against a match pattern requiring uppercase, an uppercase value that is stored lowercased persists a value the pattern rejects. The fix case-folds the value at the start of apply_constraints/2, so the constraints are checked against the form that is actually stored. This issue affects ash: from 1.29.0-rc0 before 3.32.2.

Fedify is a TypeScript library for building federated server apps powered by ActivityPub. Prior to versions 1.9.11, 1.10.10, 2.0.18, 2.1.14, and 2.2.3, an attacker can make use of JSON-LD features to restructure a JSON-LD document that would change how Fedify interprets it without changing its Linked Data Signature, allowing them to alter a third-party signed activity they have received. Versions 1.9.11, 1.10.10, 2.0.18, 2.1.14, and 2.2.3 fix the issue.

go-git is an extensible git implementation library written in pure Go. Prior to 5.19.0 and 6.0.0-alpha.3, go-git may parse malformed Git objects in a way that differs from upstream Git. When commit or tag objects contain ambiguous or malformed headers, go-git’s decoded representation may expose values differently from how Git itself would interpret or reject the same object. Additionally, go-git’s commit signing and verification logic operates over commit data reconstructed from go-git’s parsed representation rather than the original raw object bytes. As a result, go-git may sign or verify a commit payload that is not byte-for-byte equivalent to the object stored in the repository. This can cause a signature to appear valid for a commit whose displayed or effective metadata differs from the object that was intended to be signed. This vulnerability is fixed in 5.19.0 and 6.0.0-alpha.3.

NVIDIA DGX Spark GB10 contains a vulnerability in SROOT firmware, where an attacker could cause improper processing of input data. A successful exploit of this vulnerability might lead to information disclosure or denial of service.

The ip-utils package through 2.4.0 for Node.js might allow SSRF because some IP addresses (such as 0x7f.1) are improperly categorized as globally routable via a falsy isPrivate return value.

OpenClaw is a coding agent distributed as the npm package `openclaw`. In affected versions (2026.7.1 through 2026.7.2), alternate but valid `file:` URL spellings supplied over the Agent Client Protocol (ACP) were treated as relative paths and were incorrectly classified as reads scoped to the session working directory. When an operator connected `openclaw acp client` to an untrusted or compromised ACP peer, that peer could request a read of a file outside the session working directory without the approval prompt normally required for that path, resulting in disclosure of local file contents. The demonstrated impact is limited to file confidentiality; mutating and command-capable tool classes are not affected. This issue is fixed in OpenClaw 2026.8.1.