The internet is broken right now.
44 critical CVEs without a fix in the last 48 hours. Developers are crying quietly into their IDEs. 71 more already have a fix that someone still has to apply.
-
CVE-2026-133613
Flowintel's community data launcher script passes the admin API key as a command-line argument to a child Python helper process. On Unix-like systems, process arguments are readable by all local users via standard tools such as ps or by reading /proc/<pid>/cmdline, allowing any local user to observe the key while the helper process is running. An attacker with local access to the host can capture the admin API key during the brief window in which the helper process is active. Possession of the key grants full administrative access to the Flowintel REST API, enabling unauthorized creation, modification, or deletion of threat intelligence data, users, roles, and organisations. Preconditions: - The victim must invoke the community data import or deletion command (launch.sh -tdc or -dtdc) with the API key supplied as a positional argument. - The attacker must have a local user account on the same host and be able to observe process listings during the helper execution window. Impact: - Disclosure of the Flowintel admin API key to any local user. - Subsequent full administrative control over the Flowintel instance via its REST API. Affected: <4.1.0
-
CVE-2026-133495
A weakness has been identified in D-Link DIR-600L 2.17b02. Affected by this issue is some unknown functionality of the file hedwig.cgi of the component fatlady Handler. Executing a manipulation can lead to path traversal. The attack may be launched remotely.
-
CVE-2026-133492
A vulnerability has been found in TRENDnet TEW-657BRM 1.00.1. The impacted element is an unknown function of the file /usr/www/images/.svn/entries of the component mini_httpd. The manipulation leads to information disclosure. It is possible to initiate the attack remotely. The exploit has been disclosed to the public and may be used. The vendor explains: "I am not able to confirm if such vulnerability exists as the product has been EOL for more than 15 years. We will send out an announcement letting customers know." This vulnerability only affects products that are no longer supported by the maintainer.
-
CVE-2026-133489
A security vulnerability has been detected in XiangLi1999 Diffusion-LM up to 759889d58ef38e2eed41a8c34db8032e072826f4. This issue affects the function main of the file improved-diffusion/scripts/text_sample.py. Such manipulation of the argument out_dir leads to path traversal. The attack can be executed remotely. The exploit has been disclosed publicly and may be used. This product implements a rolling release for ongoing delivery, which means version information for affected or updated releases is unavailable. The project was informed of the problem early through an issue report but has not responded yet.
-
CVE-2026-133486
A vulnerability was identified in hasan-py Hayroo up to fd0fbe763153f72bfaa04ed1e1e4a1bb6f7c61df. Affected by this issue is the function Product.deleteImages of the file server/controller/products.js of the component Image Deletion Handler. The manipulation leads to path traversal. The attack may be initiated remotely. The exploit is publicly available and might be used. This product uses a rolling release model to deliver continuous updates. As a result, specific version information for affected or updated releases is not available. The identifier of the patch is 9714947bb43ec8559287aa3962b70a7be8c8d6b1. It is recommended to apply a patch to fix this issue.
-
CVE-2026-133479
A vulnerability was detected in WebKul QloApps up to 1.8.0. Affected by this vulnerability is the function postProcessDownload/installTheme of the file controllers/admin/AdminModulesController.php of the component Archive Import. The manipulation results in path traversal. It is possible to launch the attack remotely. The exploit is now public and may be used. The vendor was contacted early about this disclosure but did not respond in any way.
-
CVE-2026-133520
Dokploy 0.28.0 before 0.31.1 contains a path traversal vulnerability in the patch.readRepoDirectories tRPC procedure that allows authenticated organization members to list tracked files of arbitrary git repositories. Attackers with service read permission can supply a crafted repoPath, which is passed to git ls-tree unconfined, to enumerate other organizations' application and compose source trees.
-
CVE-2026-77312
Flyto2 Core is an execution kernel for automation and AI-agent workflows. Prior to 2.26.8, the earlier CVE-2026-67429 remediation applied validate_path_with_env_config() to only a hand-picked set of file-writing modules, leaving data.csv.write in src/core/modules/atomic/data/csv_write.py, data.json_to_csv in src/core/modules/atomic/data/json_to_csv.py, file.copy in src/core/modules/atomic/file/copy.py, and file.move in src/core/modules/atomic/file/move.py outside the FLYTO_SANDBOX_DIR confinement boundary. The advisory also identifies unguarded modules across the broader data.*, file.*, and report-writer families. data.csv.write relies on a .. substring denylist that does not reject absolute paths, while the other named modules accept caller-controlled output or destination paths without the canonical guard. A caller able to execute modules or recipes can therefore write attacker-controlled content outside the configured sandbox, potentially overwriting application code, recipes, startup files, scheduled-task files, or authorization files and gaining code execution on the host. This issue is fixed in version 2.26.8.
-
CVE-2026-77314
Flyto2 Core is an execution kernel for automation and AI-agent workflows. Prior to 2.26.8, the MCP run_recipe tool in src/core/mcp_handler.py accepts a caller-controlled recipe_name that load_recipe() in src/cli/recipe.py appends to the src/recipes path without resolved-path containment. Parent-directory segments can therefore make Flyto load and execute a YAML workflow outside the bundled recipe directory, bypassing the boundary represented by list_recipes(). This does not necessarily provide unrestricted code execution because available workflow modules remain subject to Flyto's module policy, but it permits unintended execution of allowed browser, network, file, and data-processing modules. This issue is fixed in version 2.26.8.
-
CVE-2026-77315
Flyto2 Core is an execution kernel for automation and AI-agent workflows. Prior to 2.26.8, http.paginate in src/core/modules/atomic/http/paginate.py validates only the initial base_url, and _paginate_link_header then follows a server-supplied Link: rel="next" URL without calling the SSRF guard again. A low-privileged caller can make an allowed first page point the next request at an internal or metadata service, causing a readable second-order SSRF whose response body is aggregated into all_items and returned. This issue is fixed in version 2.26.8.
-
CVE-2026-77313
Flyto2 Core is an execution kernel for automation and AI-agent workflows. Prior to 2.26.8, validate_url_ssrf in src/core/utils.py resolves and checks a hostname but returns the hostname URL rather than the validated address, and session.get in http.get performs an independent DNS lookup when connecting. A low-privileged caller with control of the hostname's DNS can make the first lookup return a public address and the connection-time lookup return an internal address, bypassing the SSRF guard for all modules that call it and exposing readable internal or metadata-service responses. Successful exploitation requires attacker-controlled DNS and a successful timing-dependent answer change between the two lookups. This issue is fixed in version 2.26.8.
-
CVE-2026-77316
Flyto2 Core is an execution kernel for automation and AI-agent workflows. Prior to 2.26.8, validate_url_ssrf() in src/core/utils.py validates one DNS answer and then returns the hostname, while guarded_aiohttp_request(), enforce_outbound_url(), and validate_url_with_env_config() allow the outbound connection to perform an independent second lookup. A caller with control of the hostname's DNS can make validation see a public address and make the connection see an internal address, bypassing the guard across its outbound consumers and exposing readable internal or metadata-service responses. The attack requires low or zero DNS TTL behavior and a successful timing-dependent answer change; static multiple-address answers are checked and do not provide the same bypass. This issue is fixed in version 2.26.8.
-
CVE-2026-61798
netty-incubator-codec-ohttp implements Oblivious HTTP (OHTTP) gateway and client functionality using Netty. Prior to 0.0.23.Final, BoringSSLAsymmetricCipherKeyPair.toString in codec-ohttp-hpke-classes-boringssl/src/main/java/io/netty/incubator/codec/hpke/boringssl/BoringSSLAsymmetricCipherKeyPair.java includes its private parameter, BoringSSLAsymmetricKeyParameter.toString in codec-ohttp-hpke-classes-boringssl/src/main/java/io/netty/incubator/codec/hpke/boringssl/BoringSSLAsymmetricKeyParameter.java renders the complete private-key byte array when isPrivate is true, and BoringSSL.EVP_HPKE_KEY_init_or_throw in codec-ohttp-hpke-classes-boringssl/src/main/java/io/netty/incubator/codec/hpke/boringssl/BoringSSL.java includes rejected privateKeyBytes in an exception message. Applications that log affected key-pair objects or initialization exceptions can persist complete HPKE private key material in logs or telemetry. Anyone who can read those records can recover the key and may compromise OHTTP message confidentiality or authenticity expectations while the key remains usable. This issue is fixed in version 0.0.23.Final.
-
CVE-2026-114516
Etherpad is a real-time collaborative editor. From 1.8.14 until 3.3.3, the pre-authentication /static/* plugin asset handler in src/node/utils/Minify.ts performs a backslash-to-forward-slash conversion after src/node/utils/sanitizePathname.ts has completed its POSIX path checks. On POSIX systems the sanitizer correctly treats backslashes as ordinary filename bytes, but the later conversion turns encoded backslash-separated parent-directory segments into traversal components without revalidation or containment. The expressPreSession registration and the always-present ep_etherpad-lite core pseudo-plugin make this path reachable without credentials, allowing a remote attacker to read any file available to the Etherpad process. Disclosed configuration or environment secrets can expose administrator credentials, after which the administrator plugin installer can provide in-process code execution and full server compromise. This issue is fixed in version 3.3.3.
-
CVE-2026-84273
IBM Guardium Data Protection 12.2 is vulnerable to path traversal in the datasource functionality. An authenticated attacker could exploit this vulnerability to delete arbitrary files on the appliance.
-
CVE-2026-92907
IBM watsonx.data 2.4.0 through 2.4.0 patch 8 could allow a remote authenticated attacker to conduct a server-side request forgery due to a server-side request forgery (SSRF) vulnerability.
-
CVE-2026-62955
baserCMS is a website development framework. From 5.2.3 until 5.3.0, BcThemeFileService::getFullpath() in plugins/bc-theme-file/src/Service/BcThemeFileService.php skips its path containment check when realpath() fails for a nonexistent theme subdirectory selected by the type parameter. An authenticated administrator can supply a traversal path through the theme-file API to access locations outside the theme directory. The bypass of the CVE-2026-30940 fix can permit file writes, overwrites or deletion within the PHP process permissions and, where executable PHP can be written, remote code execution. This issue is fixed in 5.3.0.
-
CVE-2026-130233
The Okta .NET SDK Abstractions library improperly synchronizes authentication credentials set on a shared HTTP client during request execution. When concurrent requests are dispatched for different principals, one principal's credential may be transmitted on an outbound request intended for another principal, which can result in credential exposure and lead to unauthorized actions performed as the affected principal at the Okta Management API.
-
CVE-2026-130234
The Okta Swift Management SDK does not percent-encode forward slash (/) characters during path parameter interpolation. If an application supplies input containing relative path sequences (e.g., ../) to an SDK method, standard URL normalization may direct the outbound request to an alternative Okta Management API endpoint.
-
CVE-2026-130235
In the Okta .NET SDK Abstractions library, outbound HTTP requests are not restricted to the configured Okta organization domain when processing certain URL formats. If a URL parameter is modified to specify an external host, the library may transmit the configured SSWS API token to that unintended destination.
-
CVE-2026-129931
A component of the MongoDB Java Driver that processes connection configuration places credential values taken from that configuration into generated text and error output without replacing them with a placeholder, unlike other parts of the driver. A party able to read output produced by an application that embeds the driver may obtain database or proxy credentials in clear text.
-
CVE-2026-76532
In Splunk Enterprise versions below 10.6.1, 10.4.5, 10.2.9, 10.0.12, and 9.4.17, a user who does not hold the "admin" or "power" Splunk roles could view credentials configured for email notifications in a saved search shared with that user. The vulnerability is possible when email notifications use authentication credentials because the Saved Searches Representational State Transfer (REST) API returns those credentials in clear text. For more information see Search endpoint descriptions (https://help.splunk.com/en/splunk-enterprise/leverage-rest-apis/rest-api-reference/10.6/search-endpoints/search-endpoint-descriptions), Manage knowledge object permissions (https://help.splunk.com/en/splunk-enterprise/manage-knowledge-objects/knowledge-management-manual/10.6/get-started-with-knowledge-objects/manage-knowledge-object-permissions), and Email notification action (https://help.splunk.com/en/splunk-enterprise/alert-and-respond/alerting-manual/10.6/configure-alert-actions/email-notification-action) in the Splunk documentation.
-
CVE-2026-131299
In Splunk App for Stream versions below 8.1.7, an unauthenticated user who can reach the Splunk App for Stream Representational State Transfer (REST) API on the management port (8089) could upload an Extensible Markup Language (XML) vocabulary file, which lets the vocabulary upload endpoint read local files accessible to the Splunk process or send outbound requests. The vulnerability is possible because the vocabulary upload endpoint parses uploaded XML without an application-level policy that blocks external entity resolution. For more information see About Splunk Stream (https://help.splunk.com/en/splunk-enterprise/collect-stream-data/install-and-configure-splunk-stream/8.1/introduction/about-splunk-stream) and restmap.conf (https://help.splunk.com/en/splunk-enterprise/administer/admin-manual/9.4/configuration-file-reference/9.4.3-configuration-file-reference/restmap.conf) in the Splunk documentation.
-
CVE-2026-131300
In Splunk App for Stream versions below 8.1.7, a user who can read splunkd_access.log or the _internal Splunk index could recover the Stream forwarder authentication key and use it to impersonate a trusted Stream forwarder. The vulnerability is possible because Splunk App for Stream sends the authentication key in an internal request URL, and Splunk records that URL in splunkd_access.log. For more information see About Splunk Stream (https://help.splunk.com/en/splunk-enterprise/collect-stream-data/install-and-configure-splunk-stream/8.1/introduction/about-splunk-stream) and restmap.conf (https://help.splunk.com/en/splunk-enterprise/administer/admin-manual/9.4/configuration-file-reference/9.4.3-configuration-file-reference/restmap.conf) in the Splunk documentation.
-
CVE-2026-131277
In Splunk SOAR versions below 8.8.0, a user that holds a role with permission to edit playbooks and repositories could store a modified utility block in a playbook. If a user that holds a role with permission to edit playbook code later opens and saves the playbook in the Visual Playbook Editor, a subsequent playbook run could send a Representational State Transfer (REST) API delete request to a resource outside the intended data list. The user account that runs the playbook must have permission to delete the target resource. The vulnerability is possible because Splunk SOAR does not prevent a utility block from changing the intended REST API request path. For more information see Data management automation API (https://help.splunk.com/en/splunk-soar/soar-on-premises/python-playbook-api-reference/6.3.0/automation-api/data-management-automation-api) and Manage roles and permissions in Splunk SOAR (On-premises) (https://help.splunk.com/en/splunk-soar/soar-on-premises/administer-soar-on-premises/8.8.0/manage-your-splunk-soar-on-premises-users-and-accounts/manage-roles-and-permissions-in-splunk-soar-on-premises) in the Splunk documentation.
-
CVE-2026-131282
In Splunk SOAR versions below 8.8.0, a user who holds the "Observer" Splunk SOAR role could access a password-reset credential in a diagnostic log bundle and reset the password for an active local account. The vulnerability is possible because Splunk SOAR does not redact password-reset email bodies before it logs communications between product processes.
-
CVE-2026-131266
In Splunk SOAR versions below 8.8.0, a user that holds a role with the System Settings edit permission could cause Splunk SOAR to send a stored Elasticsearch password to a host that the user controls. The vulnerability is possible because Splunk SOAR reuses a stored Elasticsearch password for a changed Search Settings endpoint without requiring a replacement password.
-
CVE-2026-131270
In Splunk SOAR versions below 8.8.0, an authenticated user with repository edit access could cause Splunk SOAR to write files outside the configured repository, which lets the user modify files that the user account running Splunk SOAR can write to. The vulnerability is possible because Splunk SOAR does not require source control repository paths to remain inside the configured repository.
-
CVE-2026-131272
In Splunk SOAR versions below 8.8.0, a user that holds a role with the System Settings edit permission could use the Automation Dependency Wheel Representational State Transfer (REST) API to write arbitrary files outside the intended directory, which could lead to arbitrary code execution. The vulnerability is possible because Splunk SOAR does not restrict uploaded automation dependency wheel filenames to the intended storage directory.
-
CVE-2026-131243
In Splunk SOAR versions below 8.8.0, an authenticated user with permission to view an affected container could query the Representational State Transfer (REST) API for approvals to retrieve a live external prompt link belonging to another user. Using this link, the user could read the approval prompt without authentication and submit a response, potentially disrupting or altering the automated workflow governed by that approval. The vulnerability is possible because the REST API for approvals exposes the live external prompt link without verifying that the requester owns the approval.
-
CVE-2026-131245
In Splunk SOAR versions below 8.8.0, a user with edit permission for the container label of a pending external prompt could obtain the external prompt link through the Representational State Transfer (REST) API and use the link to view and resolve the associated approval. The vulnerability is possible because the approval detail REST API includes the external prompt link in responses outside the assigned approval scope. For more information see REST Approval (https://help.splunk.com/en/splunk-soar/soar-on-premises/rest-api-reference/6.3.1/approval-endpoints/rest-approval), REST External Prompt (https://help.splunk.com/en/splunk-soar/soar-on-premises/rest-api-reference/7.1.0/external-prompt-endpoints/rest-external-prompt), and Manage roles and permissions in Splunk SOAR (On-premises) (https://help.splunk.com/en/splunk-soar/soar-on-premises/administer-soar-on-premises/8.5.0/manage-your-splunk-soar-on-premises-users-and-accounts/manage-roles-and-permissions-in-splunk-soar-on-premises) in the Splunk documentation.
-
CVE-2026-131246
In Splunk SOAR versions below 8.8.0, an authenticated user with edit permission for the container label on a pending approval could read an external-prompt capability through an Approval details Representational State Transfer (REST) API response and use it to view the prompt content, change the pending approval decision, and modify its action parameters. The vulnerability is possible because an Approval details response exposes the external-prompt capability without checking approval ownership. For more information see REST External Prompt (https://help.splunk.com/en/splunk-soar/soar-on-premises/rest-api-reference/8.5.0/external-prompt-endpoints/rest-external-prompt) and Manage roles and permissions in Splunk SOAR (On-premises) (https://help.splunk.com/en/splunk-soar/soar-on-premises/administer-soar-on-premises/8.5.0/manage-your-splunk-soar-on-premises-users-and-accounts/manage-roles-and-permissions-in-splunk-soar-on-premises) in the Splunk documentation.
-
CVE-2026-131202
In Splunk Enterprise Security versions below 8.7.0, a user who holds the "ess_admin" Splunk Enterprise Security role could execute arbitrary code on the search head as the user account running Splunk Enterprise by combining lookup file upload and Configuration Checker functionality. The vulnerability is possible because Splunk Enterprise Security does not restrict lookup upload file types or keep Configuration Checker module paths within the intended directory. For more information see Users and roles for Splunk Enterprise Security (https://help.splunk.com/en/splunk-enterprise-security-8/install/8.7/installation/users-and-roles-for-splunk-enterprise-security) and Create and manage lookups in Splunk Enterprise Security (https://help.splunk.com/en/splunk-enterprise-security-8/administer/8.7/managing-security-content/create-and-manage-lookups-in-splunk-enterprise-security) in the Splunk documentation.
-
CVE-2026-131191
In Splunk Enterprise versions below 10.4.5, 10.2.9, 10.0.12, and 9.4.17, a user who does not hold the "admin" or "power" Splunk roles could read private index configuration values for other users through the deployment server Representational State Transfer (REST) API. The deployment server REST API does not limit a supplied repository location to configured deployment repositories before it reads indexes.conf stanzas.
-
CVE-2026-131186
In Splunk Enterprise versions below 10.4.5, 10.2.9, 10.0.12, and 9.4.17, a user that holds a role with the list_deployment_server capability could access sensitive information from inputs.conf files outside the intended deployment app directory through the Representational State Transfer (REST) API. The vulnerability is possible because the REST API does not restrict a deployment app name to an existing deployment app before reading input configurations. For more information see authorize.conf (https://help.splunk.com/en/splunk-enterprise/administer/admin-manual/10.4/configuration-file-reference/10.4.4-configuration-file-reference/authorize.conf), Deployment server architecture (https://help.splunk.com/en/splunk-enterprise/administer/update-your-deployment/9.0/deployment-server-and-forwarder-management/deployment-server-architecture), and inputs.conf (https://help.splunk.com/en/splunk-enterprise/administer/admin-manual/9.4/configuration-file-reference/9.4.16-configuration-file-reference/inputs.conf) in the Splunk documentation.
-
CVE-2026-130161
Tamari is a fully-featured recipe manager web application built using Python and the Flask Framework. Prior to 1.7.2, an authenticated user can bypass the recipe-autofill SSRF check by submitting a hostname under the user's control through the Autofill from URL feature; the hostname can resolve to a public address during validation and a private or internal address during the later connection. In app/myrecipes/routes.py, is_safe_autofill_url validates the first DNS response, but recipe autofill passes the original URL to requests.get, which independently resolves the hostname and can connect to the internal address. The fetched response is parsed into recipe fields, enabling internal-service reconnaissance and potential data exposure; no redirect is required, and disabling redirects does not stop this DNS-rebinding bypass. This issue is fixed in 1.7.2.
-
CVE-2026-129939
The MongoDB C Driver places a submitted connection-string option value into a diagnostic message without redaction when that option fails validation. The message is returned to the calling application and is also written to the process's standard error by the driver's default logging behavior. A party that can influence the connection string given to an application using the driver may cause that field's contents to be recorded, which may result in unintended disclosure of information the application placed in that field.
-
CVE-2026-129934
The MongoDB Go driver's connection string parser includes the raw, undecoded value of a query option in the error it returns when that value contains invalid percent encodings. When the option holds a secret, such as a TLS key file password or an AWS session token, the secret appears in plain text in the error. A party with low-privileged read access to wherever the application records or displays that error may learn the secret.
-
CVE-2026-129936
A MongoDB C# Driver component that interprets connection configuration supplied by the hosting application can include the caller-supplied configuration value in the text of the error it raises when that value cannot be interpreted. The redaction applied to that text does not cover all forms in which secret material may appear, so authentication material embedded in the configuration value can be preserved in the resulting error text and in anything that records it. Exploitation requires only the low privilege needed to observe the application's own error output or logs on the system where the driver runs; no database credentials or server connectivity are required. The result is disclosure of sensitive information to parties whose access to error output is broader than their intended access to the configuration secret.
-
CVE-2026-129928
The MongoDB PHP extension includes the full connection string it was given in the error it raises when that string cannot be parsed. A party able to influence any part of the connection string, including a part that holds no secret, can cause the resulting message to carry credentials into the application's standard error output and log files. This may result in unintended disclosure of authentication material to anyone able to read that output.
-
CVE-2026-129930
A weakness in the MongoDB Node.js driver can place sensitive information taken from an application-supplied connection string into the text of an error produced while that connection string is being parsed. The error is generated inside the application process before any database connection is made, so no database account and no network access are needed for it to occur. A party able to read the application's error output or log files may thereby obtain database credentials that the application's configuration source would otherwise keep to a narrower audience.
-
CVE-2026-129924
The MongoDB Ruby driver can include unmasked connection string content in the error it raises when a connection string cannot be parsed. If that connection string carries a secret, the secret may be disclosed in cleartext through the application's error output or logs. No privileges on the driver and no network access are needed to produce the error; a party able to read the application's error output may obtain the secret.
-
CVE-2026-129927
A credential-handling weakness in the MongoDB Python Driver allows a configuration option value that protects private key material to be written in clear text into the embedding application's own output and log records when a driver object is converted to text. A party able to read that output may obtain the credential. Obtaining it requires read access to the application's local output or log files.
-
CVE-2026-104099
Fabio is an HTTP(S) and TCP router for deploying applications managed by consul. Prior to 1.7.3, route/table.go selects targets using raw req.URL.Path, while proxy/http_proxy.go applies per-route ACL and basic-auth checks to the selected target and forwards the raw path to the origin. Dot-segment, encoded-dot, or repeated-slash variants can fall through to a less restrictive route that reaches the same backend; the backend then normalizes the path and serves the protected resource without the configured access controls. A less restrictive fallback route must reach the same backend; without one, the request returns 404. This issue is fixed in version 1.7.3.
-
CVE-2026-84322
WatchGuard Dimension returns unredacted session identifiers for logged-in users in a management API response. A low-privileged Dimension Administrator can retrieve the active session IDs of other administrators, including Super Administrators, while they are logged in, enabling account takeover.
-
CVE-2026-86094
WatchGuard Dimension records unredacted session identifiers for logged-in users in its web UI diagnostic log. A low-privileged Dimension Administrator can retrieve this log and extract a Super Administrator's session token while that administrator is logged in, enabling account takeover. This vulnerability is a secondary unmitigated attack path for CVE-2026-78174.
-
CVE-2026-81430
A vulnerability in WatchGuard Dimension's system upgrade process allows an authenticated administrator to create or overwrite arbitrary files on the appliance filesystem by uploading a specially crafted upgrade archive.
-
CVE-2026-130886
@fastify/reply-from is a Fastify plugin that forwards requests to an upstream HTTP server. In versions 12.8.1 through 12.8.3, its buildURL function checked the attacker-supplied source string for parent-directory segments, but when the source begins with the same special scheme as the base, such as http or https, the WHATWG URL parser drops the scheme and resolves the remainder as a relative path. A payload that glues the scheme to a percent-encoded parent-directory sequence therefore passes the check, because the inspected segment reads as the scheme plus the dots rather than a bare parent-directory token, while the proxy forwards the encoded traversal unchanged. An upstream that percent-decodes the request path before resolving dot-segments then serves a path outside the configured prefix, so an unauthenticated client scoped to an upstream prefix, or relying on Fastify route guards in front of the proxy, can escape that prefix and reach sibling or parent upstream endpoints. The per-segment check added in 12.8.1 is what the scheme prefix bypasses, and releases up to 12.8.0 already rejected the decoded form. The issue is fixed in @fastify/reply-from 12.8.4, and users should upgrade to 12.8.4 or later.
-
CVE-2026-130159
RepoManager is a self-hosted deb/rpm package repository. Prior to 5.13.3, the authenticated get-unit-log action accepts a caller-controlled logfile value. Validate::string() escapes HTML but leaves path traversal components unchanged. The action joins the caller-controlled logfile to SERVICE_LOGS_DIR without enforcing containment, so traversal can resolve the file read outside that directory. www/controllers/ajax/status/service.php reads and returns the resolved file, and any authenticated user can invoke the action to retrieve files readable by the www-data service account, including application configuration, source code, and the GPG signing-key passphrase. Fixed in 5.13.3.
-
CVE-2026-128137
NLTK (Natural Language Toolkit) is a suite of open source Python modules, data sets, and tutorials supporting research and development in Natural Language Processing. From 3.9.4 until 3.10.4, NLTK corpus-reader path pointers and view methods validate a requested relative path without consistently resolving intermediate directory symlinks against the reader's scoped root. A local attacker who can place such a symlink can make abspaths or words reach a readable file outside the corpus root and disclose its contents. On releases with pathsec enforcement, the file must remain inside a global NLTK allowed root; version 3.9.4 has enforcement disabled and permits a broader escape. This issue is fixed in version 3.10.4.