The internet is broken right now.
45 critical CVEs without a fix in the last 48 hours. Somewhere a staging server is public. It always is. 50 more already have a fix that someone still has to apply.
-
CVE-2026-112171
An attacker can inject malicious SQL statements through user-controlled input, allowing unauthorized access to, modification of, or disclosure of database data.
-
CVE-2026-117681
A flaw was found in konflux-ci. This vulnerability allows privilege escalation when the user interface proxy automatically disables Transport Layer Security (TLS) certificate verification for connections to internal backend services in environments lacking a default service certificate authority. Because the proxy attaches a cluster-scoped authentication token with user impersonation rights to these requests, an adjacent network attacker capable of intercepting traffic can capture the credentials to gain full administrative control of the cluster.
-
CVE-2026-115413
Missing token audience validation in the OAuth authentication mode of AWS API MCP Server before version 1.6.0 might allow local users to run AWS operations with the server's configured AWS credentials via an access token issued to another application by the same identity provider. To remediate this issue, users should upgrade to version 1.6.0 and configure the expected token audience or client ID for the server.
-
CVE-2026-115079
Vulnerability in the Oracle Hyperion Infrastructure Technology product of Oracle Hyperion (component: Installation and Configuration). The supported version that is affected is 11.2.26.0.000. Difficult to exploit vulnerability allows low privileged attacker with logon to the infrastructure where Oracle Hyperion Infrastructure Technology executes to compromise Oracle Hyperion Infrastructure Technology. Successful attacks of this vulnerability can result in unauthorized ability to cause a hang or frequently repeatable crash (complete DOS) of Oracle Hyperion Infrastructure Technology. CVSS 3.1 Base Score 4.7 (Availability impacts). CVSS Vector: (CVSS:3.1/AV:L/AC:H/PR:L/UI:N/S:U/C:N/I:N/A:H).
-
CVE-2026-116250
In server mode pgAdmin 4 validates a path a user gives for a file in their storage directory with check_access_permission() and then opened it by name. Several places opened it with a plain open() or handed the path to a utility that opened it later: the servers.json export and import and the import file listing, and the files given to pg_dump, pg_dumpall, pg_restore and psql by Backup, Restore and Import/Export Data. A symbolic link swapped in at the file or at any parent directory between the check and the open redirected the operation outside the storage directory, so a file could be written, overwritten or read, or loaded into a table, as the operating-system account running pgAdmin. A hard link to an outside file, which O_NOFOLLOW does not detect, and a FIFO, which could hang a request, had the same effect. CVE-2026-86861 fixed this for the File Manager's save_file; these were the remaining paths. Exploitation requires creating such a link or FIFO inside the user's storage directory. pgAdmin offers no way to do so, and the per-user storage directories belong to the pgAdmin account, so this requires write access to the storage backend as a different operating-system user, for example a shared or network-mounted storage directory, and winning the race. The fix opens each file by walking down from the storage directory one component at a time with O_NOFOLLOW, accepting only a regular file with one link, and passes the utilities the open descriptor (/dev/fd/N) on Linux. Error messages no longer include the server path. This issue affects pgAdmin 4: before 9.19.
-
CVE-2026-110330
A flaw was found in abrt. abrt-server trusts a client-supplied PID when creating problem dumps over abrt.socket and, running as root, copies that process's /proc data (including environ) into a dump directory owned by the connecting unprivileged user. This allows a local attacker to read environment variables and related metadata of other users' or root processes when abrtd is running.
-
CVE-2026-113957
An attacker who controls the registry, a mirror, or the network path to it may be able to serve arbitrary image content in response to a pull by digest, so that a user runs an image other than the one they pinned. The fix compares the returned digest to the pinned digest and fails the pull on a mismatch. Tag pulls are exempt, because a tag is mutable by design. The fix also verifies the hash of manifests and blobs when they are fetched. This vulnerability is addressed in container version 1.6.0 and containerization version 0.52.1.
-
CVE-2026-77055
boneIO black_debian_images gives the boneio service account excessive privileges, including full passwordless sudo access and membership in the docker group. A local attacker who compromises the boneio user can use these privileges to escalate to root. This issue was fixed in version 1.6.0: the image removes the sudoers rule and the docker group membership when it is sealed, and devices already in the field have both removed by migrations shipped with boneio 1.6.0 and applied on first start after the upgrade. Privileged operations now go through helpers that accept only a fixed set of commands, with migrations verified against Ed25519 signatures.
-
CVE-2026-123510
Vikunja before 2.7.1 fails to re-anchor PluginsDir and LogPath to service.rootpath, allowing plugins to load from and logs to be written in the startup working directory. Local attackers able to write to that directory can plant plugins loaded into the server when plugins are enabled, or read misplaced logs including audit.log.
-
CVE-2026-17137
IBM API Connect 12.1.0.0 through 12.1.1.2 could allow a local attacker to obtain sensitive information due to improper use of cryptography. The encryption applied to stored data is not implemented with sufficient strength, which weakens its protection.
-
CVE-2026-131657
Tanium addressed an incorrect privilege assignment vulnerability.
-
CVE-2026-126823
A local user with an ordinary account on the same host can place a file in a shared temporary location that the JDBC driver's native library loading routine will accept and load without verifying its origin. Loading the driver class in a Java application may then run code of that user's choosing inside the application process, which may result in disclosure or alteration of data held by that process or in the process becoming unavailable.
-
CVE-2026-131685
Tanium addressed an improper signature validation vulnerability.
-
CVE-2026-126611
Progressive Robot hMailServer 6.3.4 through 6.3.7, with a MySQL or MariaDB database, TlsMode=verify-full and no TlsCaFile, does not verify the database server's certificate's issuer or name when the connection is one that MariaDB Connector/C treats as local - Server= set to 127.0.0.1, ::1 or localhost (on Linux a Unix socket), to . on Windows (a named pipe), or to a comma-separated list of servers that holds such an element - which allows a local attacker who can stand in for the database server, for example by listening on its port while it is stopped or restarting, to receive the database sign-in, the password included, and to read and modify the traffic between hMailServer and its database. hMailServer applied verify-full by turning the client's certificate verification on without a CA file, but MariaDB Connector/C 3.4.1 and later, on a connection they treat as local and with no CA file, accept any certificate within its validity period that is self-signed or sent with its own root, for any name (Connector/C 3.4.2 to 3.4.10 also one whose issuer they cannot find), so the client accepted the certificate an impostor made, asked no proof of the password and allowed a switch to an authentication method that sends the password unhashed, while hmailserver --check-database said the certificate had been checked against the system's trust store and the application log that it had been checked. The Windows builds bundle Connector/C 3.4.9 (6.3.4 to 6.3.6) and 3.4.11 (6.3.7); on Linux the distribution's client is used, and Connector/C 3.3 and 3.4.0 and Oracle's MySQL client are not affected. The configuration is not the default: TlsMode defaults to prefer, which checks no certificate.
-
CVE-2026-92199
Quest Foglight for Databases EncryptionUtil Hardcoded Encyption Key Local Privilege Escalation Vulnerability. This vulnerability allows local attackers to escalate privileges on affected installations of Quest Foglight for Databases. An attacker must first obtain the ability to execute low-privileged code on the target system in order to exploit this vulnerability. The specific flaw exists within the symmetricDecrypt method. The issue results from the use of a hard-coded cryptographic key. An attacker can leverage this vulnerability to escalate privileges and execute arbitrary code in the context of SYSTEM. Was ZDI-CAN-29967.
-
CVE-2026-58985
In handleConnectInternal of RpcServerTrusty.cpp, there is a possible denial of service due to a logic error in the code. This could lead to local denial of service with no additional execution privileges needed. User interaction is not needed for exploitation.
-
CVE-2026-129175
U.S Department of the Interior Unstable Slope Management Program allows an unauthenticated user to inject unauthorized SQL commands that allow the attacker to gain root access and complete control over the database. Fixed on or around 2026-10-19.
-
CVE-2026-114499
codebase-memory-mcp provides local code intelligence for developers and AI agents through a persistent graph and MCP tools. Prior to 0.10.0, the indexer's .gitignore matcher implemented * and ** through recursive backtracking without a work bound. A repository containing a pattern with repeated wildcard groups and a long path could make matching work grow exponentially, allowing a single ignore rule to stall discovery indefinitely and, before worker isolation, stall the server. The impact is limited to availability, and the bounded matcher does not claim that every expensive input is eliminated. This issue is fixed in version 0.10.0.
-
CVE-2026-114496
codebase-memory-mcp provides local code intelligence for developers and AI agents through a persistent graph and MCP tools. Prior to 0.9.0, the detect_changes tool validated the base_branch argument for shell metacharacters but omitted the leading-dash check before placing the value in the revision position of a git diff command. Git therefore interpreted an option-shaped base_branch value, including --output, as a command option instead of a revision, allowing a caller to create or overwrite a caller-chosen file with Git diff output under the server process's permissions. Arbitrary code execution was not demonstrated. This issue is fixed in version 0.9.0.
-
CVE-2026-77383
mcp-ssh-manager is an MCP server for managing servers over SSH with human approval, an encrypted vault, and a live audit trail. Prior to 3.6.7, ssh_db_list, ssh_db_dump, ssh_db_import, and ssh_db_restore in src/database-manager.js interpolate caller-controlled database, table, collection, file path, user, host, port, and password values into shell-evaluated command strings without neutralization. Shell metacharacters in these arguments therefore execute arbitrary commands on the configured SSH target as the account used by the MCP server. ssh_db_list is classified as read-only and remains available in readonly and restricted server security modes, so this path bypasses the command-execution boundary those modes are intended to enforce. The dump and import builders are blocked in readonly mode but remain vulnerable in the default unrestricted mode. The earlier version 3.6.5 hardening covered only ssh_db_query query text; it did not protect the list, dump, import, or restore builders, or the connection flags used by query builders. This issue is fixed in version 3.6.7.
-
CVE-2026-129995
mcp-ssh-manager is an MCP server for managing servers over SSH with human approval, an encrypted vault, and a live audit trail. Prior to 3.8.5, ssh_backup_create, ssh_backup_restore, ssh_backup_schedule, and ssh_backup_list build remote shell commands from caller-controlled values in src/backup-manager.js and inline backup handlers in src/index.js without neutralizing shell metacharacters. Crafted database, connection, backup directory, path, exclusion, schedule, or comment arguments can therefore execute arbitrary OS commands on managed remote servers as the configured SSH user. ssh_backup_schedule can persist execution through cron. ssh_backup_list does not pass through the server policy check and remains available in readonly and restricted modes, allowing command execution despite those restrictions. Tool arguments can be supplied by a malicious or compromised MCP client or steered through indirect prompt injection; fully trusted benign arguments do not trigger the issue. The earlier database-helper remediation for CVE-2026-77383 applied shell quoting to src/database-manager.js but did not cover these parallel backup builders. This issue is fixed in version 3.8.5.
-
CVE-2026-130901
Improper limitation of a pathname to a restricted directory in the file access checks in Amazon awslabs.document-loader-mcp-server 1.0.10 through 1.0.20 might allow a context-dependent actor, via crafted document content processed by an MCP client, to read files outside the intended base directory and, in 1.0.12 through 1.0.20, to write PNG slide images outside it. This occurs when DOCUMENT_BASE_DIR is not set or is empty and the CI, GITHUB_ACTIONS or PYTEST_CURRENT_TEST environment variable is present. Versions before 1.0.10 do not restrict file access to a base directory. To remediate this issue, users should upgrade to version 1.0.21 or later. Users who cannot upgrade immediately can set DOCUMENT_BASE_DIR to the directory that holds their documents. Do not set it to /, which removes the restriction.
-
CVE-2026-129054
Insertion of Sensitive Information Into Sent Data vulnerability in Datasolution AcyMailing SMTP Newsletter acymailing allows Retrieve Embedded Sensitive Data.This issue affects AcyMailing SMTP Newsletter: from n/a through 11.1.0.
-
CVE-2026-106075
Rocket.Chat's POST /api/v1/chat.sendMessage endpoint improperly validates nested message properties, allowing authenticated users with standard privileges to inject arbitrary message metadata that is persisted and trusted by the server and clients. This can lead to persistent room-level denial of service, stored HTML/iframe injection, unauthorized modification of other users' private starred messages, permission bypass and attribution spoofing for pinned messages, forged edit attribution, and invisible user mentions.
-
CVE-2026-62320
Netatalk is a Free and Open Source file server suite for Unix-like operating systems. From 1.5.0 until 4.5.1, dsi_tcp_open in libatalk/dsi/dsi_tcp.c silently limits the bytes read from an unauthenticated DSI frame received on TCP/548 to server_quantum when the attacker-controlled dsi_len is larger, without rejecting the frame or draining the declared remainder. The same silent dsi_len truncation pattern exists in libatalk/dsi/dsi_stream.c. The leftover bytes remain queued on the TCP connection and the next dsi_stream_receive call interprets them as a new DSI header and payload, allowing a remote unauthenticated attacker to smuggle one or more additional attacker-controlled DSI frames into the same connection. This issue is fixed in version 4.5.1.
-
CVE-2026-62321
Netatalk is a Free and Open Source file server suite for Unix-like operating systems. From 3.1.0 until 4.5.1, the afpd Spotlight RPC fetchQueryResultsForContext: reply path in etc/afpd/spotlight_marshalling.c calls the top-level sl_pack function from sl_pack_filemeta on a shifted sub-buffer without passing the remaining capacity of the outer DSI reply buffer. An authenticated AFP client can request an amplified nested file-metadata tree so the nested packer applies MAX_SLQ_DAT relative to the shifted pointer and writes beyond struct DSI::data. The out-of-bounds write corrupts heap memory beyond the DSI reply buffer. This issue is fixed in version 4.5.1.
-
CVE-2026-62318
Netatalk is a Free and Open Source file server suite for Unix-like operating systems. From 2.0.0 until 4.5.1, the DHX2 and DHX2-PAM authentication state machine in etc/uams/uams_dhx2_pam.c permits logincont1 to be invoked again with the same FPLoginCont retID after it has released the global gcry_mpi_t Ra object. When DHX2 or DHX2PAM is enabled in the UAM list, a remote attacker without valid AFP credentials can repeat the first continuation step, causing gcry_mpi_powm to dereference the freed Ra value and gcry_mpi_release to free it again in a privileged afpd worker. The demonstrated result is a repeatable pre-authentication worker crash, and the use-after-free and double-free provide a plausible allocator exploitation primitive although code execution was not demonstrated. This issue is fixed in version 4.5.1.
-
CVE-2026-59987
Avahi is a system which facilitates service discovery on a local network via the mDNS/DNS-SD protocol suite. In 0.9-rc4 and earlier, org.freedesktop.Avahi.Server.RecordBrowserNew accepts a domain-name value that avahi_is_valid_domain_name considers valid even though avahi_normalize_name_strdup cannot normalize it. The resulting null key reaches the key assertion in dbus_prepare_record_browser_object. A local low-privileged D-Bus client can trigger the mismatch and terminate the Avahi daemon, causing denial of service. The issue is fixed in 0.9-rc5.
-
CVE-2026-45356
Netatalk is a Free and Open Source file server suite for Unix-like operating systems. Prior to 4.4.3, the afpd Spotlight RPC unmarshalling loop in etc/afpd/spotlight_marshalling.c subtracts an attacker-controlled subcount from the unsigned count variable in sl_unpack_loop without verifying that subcount is not larger. An authenticated AFP client can place an excessive element count in the sl_unpack_ints, sl_unpack_uuid, sl_unpack_floats, or sl_unpack_date handler, or in a null element, so the subtraction wraps near UINT_MAX and the loop continues reading heap memory at increasing offsets up to toc_offset, which is capped at 65000. The out-of-bounds values can be stored in dalloc structures and returned in the Spotlight RPC response, disclosing heap contents. This issue is fixed in version 4.4.3.
-
CVE-2026-49387
Netatalk is a Free and Open Source file server suite for Unix-like operating systems. From 3.1.0 until 4.5.0, the afpd FPSpotlightRPC unmarshalling handlers in etc/afpd/spotlight_marshalling.c trust the element count encoded in the upper 32 bits of an RPC tag without comparing it with the remaining input length. An authenticated client with access to a Spotlight-enabled volume can make sl_unpack_ints, sl_unpack_date, sl_unpack_uuid, or sl_unpack_floats iterate beyond dsi->commands and read heap memory. The read values remain in internal dalloc storage and no information-disclosure channel was demonstrated, but the worker serving the attacker's session can be crashed. This issue is fixed in version 4.5.0.
-
CVE-2026-49389
Netatalk is a Free and Open Source file server suite for Unix-like operating systems. From 2.0.0 until 4.5.0, afp_catsearch in etc/afpd/catsearch.c computes the spec2 pointer for FPCatSearch and FPCatSearchExt from the unvalidated 16-bit spec_len request field without checking it against the remaining input length. An authenticated AFP client can cause the CatSearch handler to read stale bytes beyond the supplied request data. The bytes are used only as internal comparison criteria and are not returned to the client, and testing under standard server quantum settings did not demonstrate a crash or information disclosure. This issue is fixed in version 4.5.0.
-
CVE-2026-49390
Netatalk is a Free and Open Source file server suite for Unix-like operating systems. From 3.0.0 until 4.5.0, getoption_int in libatalk/util/netatalk_conf.c accepts an out-of-range server quantum value from afp.conf without applying the documented fallback to the default. An administrator can unintentionally configure a very large allocation whose failure causes AFP_PANIC at startup or a very small command buffer that disrupts DSI framing, leaving afpd unavailable until the configuration is corrected and the daemon is restarted. The option is read locally at startup and cannot be changed through the AFP protocol. This issue is fixed in version 4.5.0.
-
CVE-2026-49388
Netatalk is a Free and Open Source file server suite for Unix-like operating systems. From 3.1.0 until 4.5.0, sl_unpack_loop in etc/afpd/spotlight_marshalling.c uses an attacker-controlled Spotlight table-of-contents index during FPSpotlightRPC without checking it against the number of entries or the remaining dsi->commands buffer. An authenticated AFP client with access to a Spotlight-enabled volume can choose a large toc_index whose multiplication and addition produce an out-of-range address, causing an attacker-influenced heap read and reliably crashing the worker serving that session. The earlier CVE-2026-44066 remediation bounded recursion depth and toc_offset but did not bound toc_index. This issue is fixed in version 4.5.0.
-
CVE-2026-45354
Netatalk is a Free and Open Source file server suite for Unix-like operating systems. Prior to 4.4.3, libatalk/dsi/dsi_stream.c interprets the dsi_code and dsi_doff union field as a data offset for all DSI command types instead of limiting that interpretation to DSIWrite. An unauthenticated client can send a non-WRITE DSIFUNC_CMD packet with a nonzero dsi_code smaller than dsi_len, causing dsi_stream_receive to consume only the dsi_code number of payload bytes and leave attacker-controlled bytes in the readahead buffer. The residual bytes become the next DSI header, allowing protocol desynchronization that can close sessions, corrupt protocol state, or trigger an infinite-read denial of service. This issue is fixed in version 4.4.3.
-
CVE-2026-45355
Netatalk is a Free and Open Source file server suite for Unix-like operating systems. Prior to 4.4.3, the afpd Spotlight RPC string unmarshalling path in etc/afpd/spotlight_marshalling.c uses sl_unpack_cpx to calculate slen from qlen and the attacker-controlled used_in_last_block value without preventing a negative result. An authenticated AFP client can send a complex string or UTF-16 string element that makes slen negative, and the implicit conversion to size_t causes dalloc_strndup or convert_string_allocate to read beyond the heap buffer. The out-of-bounds read can expose heap data through the RPC response or crash the worker process. This issue is fixed in version 4.4.3.
-
CVE-2026-110779
Netatalk is a Free and Open Source file server suite for Unix-like operating systems. From 3.1.0 until 4.5.2, sl_createCNIDArray in etc/afpd/spotlight.c assumes every element of kMDQueryItemArray points to an eight-byte uint64_t and copies sizeof(uint64_t) bytes from the indexed p->dd_talloc_array element without validating the element's talloc type or allocation size. An authenticated AFP client with access to a Spotlight-enabled volume can use FPSpotlightRPC to supply a mixed-type sl_array_t containing a shorter boolean, nil, or string element, and an unauthenticated client can reach the same path when the optional guest UAM and a guest-accessible volume are enabled. The resulting eight-byte read extends beyond the shorter element's heap allocation and supplies an internal CNID filter; no confidentiality channel, write primitive, or code execution was demonstrated, while a production crash would require additional heap placement conditions. The container-type validation added for CVE-2023-42464 does not validate each element inside an accepted sl_array_t and therefore does not resolve this issue. This issue is fixed in version 4.5.2.
-
CVE-2026-110780
Netatalk is a Free and Open Source file server suite for Unix-like operating systems. From 3.1.0 until 4.5.2, the Spotlight RPC consumer handlers sl_rpc_fetchAttributeNamesForOIDArray, sl_rpc_storeAttributesForOIDArray, and sl_rpc_fetchAttributesForOIDArray in etc/afpd/spotlight.c dereference the first element of a decoded sl_cnids_t without checking whether the inner CNID array is empty. An authenticated AFP client, or a guest when guest AFP access is enabled, can send a crafted SPOTLIGHT_CMD_RPC whose outer CNIDS length is greater than 16 while its inner count is zero, causing sl_unpack_cnids to create an empty DALLOC_CTX and the consumer to dereference dd_talloc_array[0]. The affected afpd child crashes, normally limiting denial of service to the attacker's session, and no memory write or code execution was demonstrated. This issue is fixed in version 4.5.2.
-
CVE-2026-103933
Netatalk is a Free and Open Source file server suite for Unix-like operating systems. From 2.1.0 until 4.5.2, get_eacontent in libatalk/vfs/ea_ad.c subtracts MAX_REPLY_EXTRA_BYTES from a client-controlled FPGetExtAttr MaxReplySize without rejecting values from 1 through 5, and a signed-to-unsigned comparison can then select an untrusted AppleDouble ea_size. An authenticated client on a volume configured with ea = ad can use a pre-existing malformed AppleDouble sidecar to make read copy attacker-controlled bytes beyond dsi->data + 6 into the surrounding struct DSI, including protocol callback pointers. The malformed sidecar must already exist or be created through a separate local, co-export, management, restore, or vulnerability-assisted write path, and practical code execution depends on the target architecture, structure layout, and address uncertainty; runtime command execution was demonstrated on one x86-64 build. Volumes using ea = sys, ea = none, or ea = auto when it resolves to native system xattrs do not reach this path. This issue is fixed in version 4.5.2.
-
CVE-2026-133879
A flaw was found in KubeVirt/containerized-data-importer. When a VirtualMachine with an embedded DataVolumeTemplate carrying the cdi.kubevirt.io/storage.checkStaticVolume annotation and a cross-namespace clone source is created by a tenant, the service-account clone authorization check can be bypassed, allowing a PVC from another namespace to be cloned into the tenant's namespace without the required permissions. This can result in cross-tenant data disclosure.
-
CVE-2026-133872
LessPass self-hosted sync server backend before commit 9469b0c lacks rate limiting on the JWT login endpoints /tokens/auth/ and /auth/jwt/create/, allowing unlimited authentication attempts. Unauthenticated remote attackers can brute force passwords or perform credential stuffing against known account emails to obtain tokens granting read and write access to stored profiles.
-
CVE-2026-133873
Cloud Mail before 3.6.0 contains a stored cross-site scripting vulnerability in attService.addAtt that allows unauthenticated attackers to store inline attachments with sender-supplied MIME types. Attackers can email SVG or HTML attachments, compute the content-hash URL under /oss/ or /attachments/, and steal session JWTs from localStorage to hijack accounts including administrators.
-
CVE-2026-133875
one-hub through 0.14.27 contains a server-side request forgery vulnerability in GetImageFromUrl in common/image/image.go that fetches vision chat image_url values without destination restrictions. Attackers holding a standard non-admin API token can send crafted /v1/chat/completions requests via Claude, OpenRouter or Ollama providers to reach internal services or cloud metadata endpoints.
-
CVE-2026-132663
In Postgres Enterprise Manager (PEM), webhook URLs are not validated, and the PEM agent does not restrict the URL schemes it uses to deliver webhooks. A user with the alert configuration role (pem_config_alert) can set up a webhook with a file:// URL. When the webhook is sent, the agent on the PEM server, which runs as root by default, reads the local file and stores its contents in the webhook error record in the PEM database, where any PEM user can read it. From PEM 10.6.1, PEM accepts and sends only http and https webhook URLs.
-
CVE-2026-132664
In Postgres Enterprise Manager (PEM), any PEM user can read the encrypted passwords that PEM agents use to connect to monitored servers, including the PEM backend superuser password stored when the PEM server is configured. Unless a custom encryption key is configured, these passwords are encrypted with a key that is the same in every PEM installation, so they can be decrypted. From PEM 10.6.1, PEM users can no longer read these passwords.
-
CVE-2026-132665
In Postgres Enterprise Manager (PEM), the pem.deploy_job_template() function does not check that the caller can access the servers, groups and agents it deploys a job template to. When PEM teams are used to separate access, a user with the scheduled tasks role (pem_manage_schedule_task) can deploy and schedule SQL or shell job steps on agents that belong to other teams. From PEM 10.6.1, PEM checks team access for every server, group and agent a template is deployed to.
-
CVE-2026-132668
In Postgres Enterprise Manager (PEM), any PEM user can run the pem.send_webhook() function, which runs with elevated privileges and queues a payload for any configured webhook. The PEM agent then delivers it with that webhook's stored headers, OAuth token or client certificate, so a user can send forged notifications to endpoints they could not otherwise authenticate to. From PEM 10.6.1, no PEM login role can call this function directly. PEM still sends webhook notifications through its own internal paths.
-
CVE-2026-132658
In Postgres Enterprise Manager (PEM), the pem_agent role, which every PEM agent login inherits, is a member of the PostgreSQL pg_read_server_files and pg_write_server_files roles on the PEM backend database. An attacker who controls any monitored host, and so its PEM agent certificate, can read and write files on the PEM server host as the postgres operating system user. This includes the PEM certificate authority key, which lets the attacker issue certificates for other agents and act as them. From PEM 10.6.1, PEM agents no longer hold these roles.
-
CVE-2026-132659
In Postgres Enterprise Manager (PEM), the row-level security policies on pem.job and pem.jobstep that apply to PEM agents do not limit an agent to its own jobs. An attacker who controls one monitored host, and so its PEM agent certificate, can create, change or delete jobs for every other agent, including the agent on the PEM server. Agents run SQL job steps with the database credentials of the server they monitor, and shell job steps where batch jobs are enabled, so compromising one monitored host can give superuser SQL access to every monitored database and to the PEM backend. From PEM 10.6.1, PEM agents can no longer create jobs or change job definitions or job steps.
-
CVE-2026-132660
In Postgres Enterprise Manager (PEM), several SECURITY DEFINER functions in the pem schema of the PEM backend database are owned by a superuser but do not set a safe search_path. Any PEM user who can connect to the PEM backend database can create objects that these functions use in place of the intended ones, and so run arbitrary SQL with superuser privileges. Temporary objects can be used for this on every supported PostgreSQL version, and objects in the public schema on PostgreSQL 14 backends and backends upgraded from PostgreSQL 14. From PEM 10.6.1, these functions run with a fixed search_path.
-
CVE-2026-127027
Poste.io is vulnerable to remote code execution in DMARC report processing. Email sent to dmarc-reports@<domain> is saved without sender verification. Script that runs as root unpacks the attachments using unsafe shell commands, allowing command injection through crafted attachments (e.g. file names or archive contents). A remote unauthenticated attacker who can send an email to the server's port 25 can execute arbitrary commands as root within the Poste.io container. This issue was fixed in version 2.5.22.