Roo Code through 3.54.0 contains a command injection vulnerability in the auto-approve execute feature that allows attackers to bypass allowlist/denylist enforcement by nesting command substitutions inside parameter expansion defaults. The command parser in parse-command.ts replaces parameter expansions with opaque placeholders before extracting command substitutions, causing the containsDangerousSubstitution guard to miss nested payloads, which are then auto-approved based on the outer allowlisted command prefix and executed by the shell via execa, enabling arbitrary command execution.

A vulnerability has been found in princezuda SafestClaw up to 4.2.4. This vulnerability affects the function ShellAction._validate_command of the file src/safestclaw/actions/shell.py of the component Built-in Web Interface. Such manipulation leads to incomplete blacklist. An attack has to be approached locally. The exploit has been disclosed to the public and may be used. The presence of this vulnerability remains uncertain at this time. The project maintainer explains: "On paper you're correct, this is a vulnerability. In practice, nothing your AI generated shows how it makes users vulnerable. It's open source. Someone can mod the shell allow list or remove that system. Present an actual poc that shows a threat to users."

A flaw was found in the Keycloak keycloak-services component, which handles the management of identity providers. The issue occurs when a delegated administrator updates an OIDC identity provider using a masked client secret sentinel value. Due to improper validation, Keycloak reuses the existing real secret even if security-sensitive fields like the token URL have been changed, allowing an attacker to redirect and capture the secret.

Improper Validation of Specified Quantity in Input in ZenHive mpp allows an unauthenticated remote client to drain the fee-payer wallet, resulting in denial of service for legitimate clients. When the mpp Elixir library is configured as fee payer (fee_payer: true), the MPP.Methods.Tempo payment method co-signs and broadcasts a client-supplied EVM transaction without first validating that the client-supplied gas_limit is sufficient to complete the intended call. A malicious client can submit a signed transferWithMemo transaction with gas_limit deliberately set just below the amount required for successful execution. The server co-signs the transaction and broadcasts it via rpc_broadcast_sync. The transaction runs out of gas during EVM execution and reverts, but the fee-payer wallet is still charged for the burned gas while the client pays nothing and receives no resource. Repeated requests from one or more malicious clients drain the fee-payer wallet at near-zero cost to the attacker, ultimately preventing the server from sponsoring gas for legitimate payment requests. The wait_for_confirmation = false (optimistic) path is also affected: it invokes simulate_payment_call via eth_call, but that simulation omits the gas parameter and therefore does not catch out-of-gas conditions. This issue affects mpp: from 0.2.0 before 0.6.0.

Improper Validation of Specified Quantity in Input in ZenHive mpp allows an unauthenticated remote client to inflate the fee-payer's gas cost per payment by a large multiplier, degrading the sponsor's operating margin. When the mpp Elixir library is configured as fee payer (fee_payer: true), MPP.Tempo.Transaction.cosign_fee_payer/3 re-signs the client-supplied base fields of the 0x76 AASigned envelope verbatim, including the EIP-2930 access list, without validating its length or contents. EIP-2930 access list entries incur intrinsic gas (~2,400 gas per address, plus 1,900 gas per storage key) charged before any opcode executes, regardless of whether the listed addresses are ever touched. A malicious client submits a valid transferWithMemo call alongside a large number of fabricated access-list entries. The server co-signs and broadcasts the transaction. The intended transfer executes normally, but the fee-payer wallet pays a large multiple of the expected gas cost with no corresponding on-chain work. At the maintainer's default of 137 access-list entries (fitting within Bandit's 10,000-byte per-header-field limit) and 100 Gwei max_fee_per_gas, per-payment gas cost rises from ~51,287 to ~380,087 gas, a 7.4x multiplier. Sustained abuse destroys the sponsor's operating margin on low-cost payments and, over time, drains the fee-payer wallet. This issue affects mpp: from 0.2.0 before 0.6.0.

Improper Validation of Specified Quantity in Input in ZenHive mpp allows an unauthenticated remote client to drain the fee-payer wallet in a single request by naming an arbitrarily high gas price. When the mpp Elixir library is configured as fee payer (fee_payer: true), MPP.Tempo.Transaction.cosign_fee_payer/3 re-signs the client-supplied base fields of the 0x76 AASigned envelope verbatim, including max_fee_per_gas and max_priority_fee_per_gas, without validating that they are within reasonable bounds. A malicious client embeds arbitrarily large values for these fields in the signed envelope. The server co-signs and broadcasts the transaction. The effective_gas_price billed against the fee-payer wallet is derived from the attacker-supplied ceilings, so the server pays those inflated per-gas rates out of its own wallet. A single crafted request can drain the wallet entirely, after which the server can no longer sponsor gas for legitimate payment requests. This issue affects mpp: from 0.2.0 before 0.6.0.

OpenClaw versions before 2026.6.6 contain an environment variable filtering vulnerability in host exec that fails to properly sanitize rustup startup variables. Attackers with lower-trust caller access or configured input paths can execute or persist actions beyond their intended authorization level.

The Apify MCP server enables AI agents to extract data from websites using ready-made scrapers, crawlers, and automation tools available on the Apify Store. Prior to 0.9.21, the fetch-apify-docs tool in src/tools/common/fetch_apify_docs.ts validates allowlisted documentation domains with String.startsWith() rather than URL hostname comparison, allowing attacker-controlled URLs such as `https://docs.apify.com.evil.com/` and `https://docs.apify.com@evil.com/` to pass the ALLOWED_DOC_DOMAINS check and return arbitrary fetched content to the LLM. This issue is fixed in version 0.9.21.

Squid is a caching proxy for the Web. Prior to 7.6, due to an improper validation of syntactic correctness of input in the FTP gateway (src/clients/FtpGateway.cc), Squid is vulnerable to an out-of-bounds read: when a listing entry date in the TypeA or TypeB directory-listing formats is not followed by a filename, parsing was not restricted to the input buffer, so a trusted client accessing a misbehaving FTP server through Squid's gateway feature could read memory from random unrelated transactions. This issue is fixed in version 7.6.

stoatchat versions before 0.7.8 fail to enforce account creation restrictions including invite-only mode, email verification, captcha, and shield verification. Attackers can create unlimited accounts with unverified email addresses, increasing denial-of-service risk and compromising service integrity.

NocoBase is an AI-powered no-code/low-code platform for building business applications and enterprise solutions. In 2.0.59 and earlier, NocoBase @nocobase/plugin-collection-sql used the checkSQL() function in packages/plugins/@nocobase/plugin-collection-sql/src/server/utils.ts with an incomplete keyword blacklist that did not restrict PostgreSQL system catalog tables such as pg_shadow, pg_roles, and pg_stat_activity, allowing an admin-role user to read password hashes and database metadata through the SQL Collection feature. This vulnerability is fixed in 2.1.0-alpha.46.

Symfony Polyfill backports PHP features and provides compatibility layers for extensions and functions. From 1.17.1 until 1.38.1, symfony/polyfill-intl-idn accepts xn-- labels whose Punycode payload is empty or decodes to ASCII-only code points because Idn::process() does not enforce the UTS #46 revision 33 requirement that decoded ACE labels contain at least one non-ASCII code point. Originally unequal domain names can be regarded as equal, which can lead to blacklist bypassing, inconsistent URL parsing, and server-side request forgery in applications using the polyfill to canonicalise or compare hostnames. This issue is fixed in version 1.38.1.

Improper validation of specified type of input in .NET Framework allows an unauthorized attacker to deny service over a network.

Symfony is a PHP framework for web and console applications and a set of reusable PHP components. From 5.4.0 to 5.4.53, 6.4.41, 7.4.13, and 8.0.13, NoPrivateNetworkHttpClient and IpUtils::PRIVATE_SUBNETS omitted IPv6 transition prefixes such as 6to4, NAT64, Teredo, and IPv4-compatible IPv6, allowing attacker-supplied URLs to represent private IPv4 targets in forms that IpUtils::isPrivateIp() did not block. This issue is fixed in versions 5.4.53, 6.4.41, 7.4.13, and 8.0.13.

Symfony is a PHP framework for web and console applications and a set of reusable PHP components. Prior to 6.4.40, 7.4.12, and 8.0.12, OidcTokenHandler::verifyClaims() registered audience (aud), issuer (iss), and expiry (exp) checkers but did not pass the mandatory claims list to ClaimCheckerManager::check(), so a validly signed JWT that omitted those claims could pass verification. This issue is fixed in versions 6.4.40, 7.4.12, and 8.0.12.

Symfony is a PHP framework for web and console applications and a set of reusable PHP components. From 6.1.0-BETA1 until 6.4.40, 7.4.12, and 8.0.12, UrlAttributeSanitizer::getSupportedAttributes() omits URL-valued attributes including action, formaction, poster, and cite, so configurations that admit those attributes can leave javascript: URIs unsanitized and enable XSS when the resulting HTML is rendered or a victim submits a form or clicks a button. This issue is fixed in versions 6.4.40, 7.4.12, and 8.0.12.

Symfony is a PHP framework for web and console applications and a set of reusable PHP components. From 6.1.0-BETA1 until 6.4.40, 7.4.12, and 8.0.12, HtmlSanitizer URL sanitization can allow off-allowlist URLs through allowLinkHosts() or allowMediaHosts() because UrlSanitizer::parse() follows RFC 3986 while browsers follow WHATWG URL parsing, and because <area href> is checked against the media policy rather than the link policy. This issue is fixed in versions 6.4.40, 7.4.12, and 8.0.12.

Improper validation of specified type of input in Microsoft Office Word allows an unauthorized attacker to disclose information locally.

Improper restriction of names for files and other resources in Github Copilot allows an unauthorized attacker to execute code locally.

Improper validation of consistency within input in Windows Secure Kernel Mode allows an authorized attacker to elevate privileges locally.

A vulnerability was found in nextlevelbuilder GoClaw 3.11.3. Affected by this issue is the function ExecApprovalManager.CheckCommand of the file internal/tools/exec_approval.go. The manipulation results in incomplete blacklist. The attack can be executed remotely. The exploit has been made public and could be used.

OpenClaw versions before 2026.6.6 contain a flaw in host exec environment filtering that could allow Git ext transport to be abused. When the affected feature is enabled and reachable, a lower-trust caller or configured input path could execute or persist actions beyond the caller's intended authorization.

OpenClaw versions before 2026.6.6 contain a flaw in host exec environment filtering that can miss interpreter startup variables. When the affected feature is enabled and reachable, a lower-trust caller or configured input path can supply crafted environment variables to execute or persist actions beyond the caller's intended authorization.

Improper Validation of Specified Quantity in Input vulnerability in WPDeveloper Better Payment – Instant Payments, Donations, Fundraising with Subscriptions &amp; More better-payment allows Accessing Functionality Not Properly Constrained by ACLs.This issue affects Better Payment – Instant Payments, Donations, Fundraising with Subscriptions &amp; More: from n/a through <= 2.2.0.

An Improper Validation of Syntactic Correctness of Input vulnerability in the SIP plugin of Juniper Networks Junos OS on MX Series with SPC3 and SRX Series allows an unauthenticated, network-based attacker to cause a Denial-of-Service (DoS).If the SIP ALG is enabled on an affected device, the processing of a malformed SIP invite packet will cause a flow processing daemon (flowd) crash and restart. This leads to a complete service outage until the system has automatically recovered. This issue affects Junos OS on MX Series with SPC3 and SRX Series: * all versions before 23.2R2-S7, * 23.4 versions before 23.4R2-S8, * 24.2 versions before 24.2R2-S5, * 24.4 versions before 24.4R2-S4, * 25.2 versions before 25.2R2, * 25.4 versions before 25.4R1-S2.

An Improper Validation of Specified Quantity in Input vulnerability in the TCP proxy plugin of Juniper Networks Junos OS on MX Series with SPC3, and SRX Series allows an unauthenticated, network-based attacker to cause a complete Denial of Service (DoS). When TCP proxy is engaged in a flow session, to support ALGs, Advanced Anti-Malware, ICAP or UTM, a TCP packet with specifically malformed TCP header will cause flow processing daemon (flowd) to crash and restart. This causes a complete service outage until the system has automatically recovered. This issue affects Junos OS on MX with SPC3, and SRX Series:  * 23.4 versions before 23.4R2-S7,  * 24.2 versions before 24.2R2-S4,  * 24.4 versions before 24.4R2-S3, * 25.2 versions before 25.2R2. This issue does not affect releases before 23.4R1.

An Improper Validation of Specified Quantity in Input vulnerability in the Packet Forwarding Engine (pfe) of Juniper Networks Junos OS on MX Series allows an unauthenticated, adjacent attacker to cause a Denial-of-Service (DoS). When a specific packet is received from device in the same broadcast domain, an affected system calculates the packet size incorrectly. This causes further packet processing to fail, which triggers an FPC major error, resulting in a FPC reset impacting traffic until the FPC has automatically recovered. Affected scenarios are: MAP-T, or non-IP traffic encapsulated in IP (e.g. MPLS over GRE). When this issue happens the following logs can be observed: fpc<#> CMError: /fpc/0/pfe/0/cm/0/MQSS(0)/0/MQSS_CMERROR_LI_INT_REG_UNROLL_TAIL_LENGTH_OVF (0x2205eb), scope: pfe, category: functional, severity: major, module: MQSS(0), type: LI: Unroll TAIL length overflow, oc_category: default fpc<#> Performing action reset-fru for error /fpc/0/pfe/0/cm/0/MQSS(0)/0/MQSS_CMERROR_LI_INT_REG_UNROLL_TAIL_LENGTH_OVF (0x2205eb) in module: MQSS(0) with scope: pfe category: functional level: major, oc_category: default This issue affects Junos OS on MX Series: * all versions before 23.2R2-S6, * 23.4 versions before 23.4R2-S7, * 24.2 versions before 24.2R2-S4, * 24.4 versions before 24.4R2-S4, * 25.2 versions before 25.2R2.

An Unchecked Input for Loop Condition vulnerability in the Packet Forwarding Engine (pfe) of Juniper Networks Junos OS on MX Series allows an unauthenticated, adjacent attacker to cause a Denial-of-Service (DoS).Micro-BFD session flaps generate respective up/down events which are queued by PFEMAN for processing. Especially in a Virtual-Chassis (VC) scenario with locality‑bias configured, processing takes a significant amount of time for each event. If these sessions keep flapping, new events are constantly added, and in turn PFEMAN never completes processing these events. This results in the PFEMAN watchdog timer expiring, which causes the FPC to crash and restart, representing a complete service outage. This issue only affects MX series FPCs up to and including MPC9, and LC2101/2103 and LC480. It does not affect MPC10/11, LC4800/9600, and MX304. This issue affects Junos OS on MX Series: * all versions before 23.2R2-S7, * 23.4 versions before 23.4R2-S8, * 24.2 versions before 24.2R2-S4, * 24.4 versions before 24.4R2-S3, * 25.2 versions before 25.2R2.

RestrictedPython is a tool that helps to define a subset of the Python language which allows to provide a program input into a trusted environment. Prior to 8.3, check_function_argument_names() rejected protected guard hook names for regular, variadic, and keyword-only arguments but omitted positional-only arguments, allowing __getattr__, _getitem_, _write_, or _print_ to be shadowed by a local parameter and bypass the embedding application's access policy. This issue is fixed in version 8.3.

FMP/NOTIFY protocol dissector crash in Wireshark 4.6.0 to 4.6.6 and 4.4.0 to 4.4.16 allows denial of service

PasswordPusher before 2.8.1 accepts data URI schemes in URL push payloads due to insufficient validation in the valid_url function. Attackers can create malicious pushes containing data:text/html URIs that execute arbitrary JavaScript in victims' browsers when clicked, enabling phishing and credential theft under the trusted PasswordPusher domain.

Mistune is a Python Markdown parser with renderers and plugins. Prior to 3.3.0, the safe_url filter in src/mistune/renderers/html.py blocks only javascript:, vbscript:, file:, and data: schemes, allowing legacy or chained schemes such as feed:, view-source:, jar:, livescript:, mocha:, ms-its:, mk:, and res: to reach rendered href and src attributes and potentially execute script in affected user agents. This issue is fixed in version 3.3.0.

Mistune is a Python Markdown parser with renderers and plugins. Prior to 3.3.0, the toc plugin and TableOfContents directive generate heading IDs as predictable toc_N values without slugifying the heading text, allowing attacker-controlled id="toc_N" content to collide with generated anchors and redirect same-page navigation, CSS selectors, or JavaScript handlers. This issue is fixed in version 3.3.0.

OpenClaw before 2026.5.28 contains a credential exposure vulnerability where workspace dotenv files can override provider credentials. Attackers with lower-trust access to configured input paths can expose sensitive data and credentials that should remain within trusted boundaries.

Immutable.js provides many Persistent Immutable data structures. Prior to 4.3.9 and 5.1.8, List#set, List#setSize, List#setIn, List#updateIn, and the functional set, setIn, and updateIn mishandle an index or size in the range 2 ** 30 to 2 ** 31 in setListBounds in src/List.js, causing an empty List to enter an uncatchable infinite loop, a populated List to allocate without bound until process abort, or setSize to silently wrap large values. This issue is fixed in versions 4.3.9 and 5.1.8.

internal-sftp in sshd in OpenSSH before 10.4 recognizes only the first 9 command-line arguments, which can be important if a later command-line argument would have helped to ensure the intended security properties of an SFTP connection.

FOSSBilling is a free, open-source billing and client management system. Prior to version 0.8.0, the PayPalEmail payment adapter accepts PayPal IPN callbacks and credits the IPN-supplied amount (`mc_gross`) to the client's balance without validating it against the invoice total. Combined with a $0.05 floating-point epsilon tolerance in the invoice credit-payment logic, this allows a client to underpay an invoice by up to $0.04 and still have it marked as fully paid. Version 0.8.0 patches the issue. There is no effective workaround without modifying the source code. Merchants using the PayPalEmail adapter should monitor IPN transactions for amounts that do not match their corresponding invoice totals, and manually review and refund suspicious payments.

vLLM is a high-throughput and memory-efficient inference and serving engine for LLMs. Prior to 0.24.0, a frontend-legal multi-request speculative decoding workload can cause the rejection sampler to produce a recovered token equal to the model vocabulary size boundary value, which is then converted to negative one when the engine selects the next live token for a request and is written back into the drafter's input ids; that out-of-vocabulary value is later consumed by the model's embedding and attention path and crashes the engine worker with a GPU device-side assertion. The same triggering request sequence is reachable through the public gRPC Generate and Abort endpoints, so a remote client that can send generation requests can crash the shared engine worker, aborting concurrent requests and causing a service-wide denial of service for other clients of the deployment until the worker is restarted. This issue is fixed in version 0.24.0.

A flaw exists in the org.keycloak.broker.oidc package where the OIDC broker incorrectly synchronizes the email_verified claim. When an OIDC identity provider is configured with trustEmail=true and the userinfo endpoint is enabled, Keycloak retrieves the email address from the userinfo response but retrieves the email_verified status exclusively from the id_token. The root cause is a lack of validation ensuring that the email_verified claim in the id_token actually refers to the email address returned by the userinfo endpoint. If these two sources return different email addresses, the id_token's email_verified=true claim is blindly applied to the userinfo email. Exploitation Conditions: The OIDC identity provider must have trustEmail set to true (non-default). The userinfo endpoint must be enabled (default). The attacker must control or have compromised the upstream OIDC provider. Concrete Impact: Mark arbitrary email addresses as verified in the Keycloak database. Bypass email-based security controls or verification workflows. Potential account takeover if the application relies solely on the email_verified flag from the IdP to link accounts.

Trail of Bits fickling versions up to and including 0.1.10 do not include the Python standard library modules _posixsubprocess, site, and atexit in the UNSAFE_IMPORTS denylist (fickle.py). Because these modules are absent from the denylist, fickling's check_safety() function returns LIKELY_SAFE with zero findings for pickle payloads that invoke dangerous functions including _posixsubprocess.fork_exec (C-level process spawner capable of executing arbitrary binaries), site.execsitecustomize (executes arbitrary site customization code), and atexit._run_exitfuncs (triggers all registered exit handler callbacks). The fickling.load() API chains check_safety() into pickle.loads() as an explicit security gate; a LIKELY_SAFE verdict causes the payload to be deserialized and executed. This shares the same root cause as CVE-2026-22607 (cProfile), CVE-2025-67748 (pty), and CVE-2025-67747 (marshal/types). OvertlyBadEvals does not flag these modules because they are standard library imports. UnsafeImports does not flag them because they are not in the denylist. The UnusedVariables heuristic is defeated by the SETITEMS opcode pattern.

** UNSUPPORTED WHEN ASSIGNED ** Improper Validation of Specified Quantity in Input in the ASUS AI Suite 3 driver allows a local user to access unintended memory regions via crafted IOCTL requests, leading to privilege escalation. Refer to the 'End-of-Life Notice and Driver Update for Legacy ASUS Drivers ' section on the ASUS Security Advisory for more information.

** UNSUPPORTED WHEN ASSIGNED ** Improper Validation of Specified Quantity in Input in the ASUS AI Suite 3 driver allows a local user to bypass security validation and access restricted memory blocks via crafted IOCTL requests, leading to privilege escalation. Refer to the 'End-of-Life Notice and Driver Update for Legacy ASUS Drivers ' section on the ASUS Security Advisory for more information.

The Erlang/OTP ssl application does not validate that the PSK identity list and binder list carried in a TLS 1.3 ClientHello pre-shared key extension have equal length before passing them to the session ticket handler. In tls_handshake_1_3:handle_pre_shared_key/3, an OfferedPreSharedKeys record with a mismatched number of identities and binders is forwarded directly to tls_server_session_ticket:use/4, which crashes the session ticket handler process. An unauthenticated remote attacker can send a single crafted ClientHello to a TLS 1.3 server with session tickets enabled (stateful or stateless mode) and permanently disrupt session ticket handling on that listener. New TLS 1.3 handshakes complete but subsequently crash when the server attempts to issue a session ticket, effectively making TLS 1.3 unusable on the affected listener until the ssl application is restarted. TLS 1.2 connections are not affected. This issue affects OTP from 22.2 before 29.0.3, 28.5.0.3 and 27.3.4.14 corresponding to ssl from 9.5 before 11.7.3, 11.6.0.3 and 11.2.12.10.

Missing validation of "valuesFrom" references in Helm Deployer of SUSE Rancher Fleet 0.15 before 0.15.2, 0.14 before 0.14.6, 0.13 before 0.13.11 and 0.12 before 0.12.15 could be used by owners of one tenant to access fleet credentials of other tenants.

Unauthenticated Arbitrary Code Execution in W3 Total Cache <= 2.9.4 versions.

n8n before 2.25.7 and 2.26.x before 2.26.2 contains an abstract syntax tree (AST) security validator bypass in the Python Code node. An authenticated user with permission to create or modify workflows containing a Python Code node can bypass the validator and access the task executor module namespace. The issue only affects self-hosted instances where the Python Task Runner is enabled; where N8N_BLOCK_RUNNER_ENV_ACCESS is configured to allow it, this can disclose environment variables accessible to the task runner process.

Picklescan before 0.0.25 fails to detect unsafe global functions in the Numpy library, allowing attackers to bypass static analysis and execute arbitrary code during deserialization. Attackers can craft malicious pickle files using numpy.testing._private.utils.runstring within the reduce method to import dangerous libraries like os and execute arbitrary OS commands when the pickle file is loaded.

IBM Db2 11.5.0 through 11.5.9, and 12.1.0 through 12.1.4 for Linux, UNIX and Windows (includes Db2 Connect Server) could allow an authenticated user to cause a denial of service due to improper neutralization of special elements in the data query logic of XMLTable-derived columns.

The ELF image activator cleared per-process ASLR preference flags for setuid binaries after the code that computes the PIE base address, rather than before. As a result, a user-requested ASLR disable was still in effect at the point where the base address was chosen. An unprivileged local user can disable ASLR for a setuid PIE binary by calling procctl(2) before execve(2). This makes exploitation of any separate memory corruption vulnerability in that binary significantly easier.

Kestra is an open-source, event-driven orchestration platform. Prior to 1.0.45 and 1.3.21, AuthenticationFilter in Kestra OSS uses request.getPath().endsWith("/configs") to whitelist the public configuration endpoint from Basic Auth. Because the check is a suffix match rather than an exact path match, any API path whose last segment is configs bypasses authentication entirely. An unauthenticated remote attacker can exploit this to create and execute arbitrary workflows without credentials. Because Kestra ships with script execution plugins (plugin-script-shell, plugin-script-python, etc.) enabled by default, this directly results in unauthenticated Remote Code Execution as root inside the Kestra worker container. This vulnerability is fixed in 1.0.45 and 1.3.21.