Valentin Lobstein

FileRun before 2026.3.0 contains a PHP object injection vulnerability that allows authenticated attackers to execute arbitrary code by exploiting incorrect options passed to unserialize() in the Perms::getPerms() method, where a positional array is used instead of the required named-key array to disable class instantiation. Attackers with database write access can inject a serialized gadget chain into the permissions table columns processed on every authenticated page load to write arbitrary files, such as PHP webshells, to web-accessible paths.

FileRun before 2026.3.0 contains a SQL injection vulnerability that allows delegated or simple administrators to execute arbitrary SQL by submitting the description parameter as an array, causing the getValuesString() method in DB/DP.php to interpolate raw array values directly into an INSERT statement without parameterization. Because the underlying PDO connection uses emulated prepared statements enabling stacked queries, attackers can manipulate the df users permissions table to escalate a delegated administrator account to superuser privileges, and may additionally achieve code execution via unsanitized path values passed to require once in the logs listing component.

FileRun before 2026.3.0 contains an OS command injection vulnerability in the PhotoProofSheet handler that allows authenticated users with upload permission to execute arbitrary commands by uploading files with shell metacharacters in their names. Attackers can upload a file containing command substitution syntax such as backticks, semicolons, or $() sequences in the filename, then trigger the PhotoProofSheet endpoint to execute arbitrary commands as the web-server user due to missing escapeshellarg() sanitization in the ImageMagick montage command construction.

FileRun before 2026.3.0 contains an OS command injection vulnerability caused by a no-op redefinition of escapeshellcmd() in CLI.php that strips shell-metacharacter escaping, allowing attacker-controlled input to reach an exec() sink unsanitized. Attackers can exploit this through an interactive path via image preview.php with a crafted args parameter requiring superuser authentication, or through a persistent path by storing malicious payloads in thumbnails ffmpeg args or thumbnails ffmpeg ss that execute when any user triggers video thumbnail generation.

**Name of the Vulnerable Software and Affected Versions** Joomla! Core versions 1.0.0 through 5.4.7 Joomla! Core versions 6.0.0 through 6.1.2 **Description** The software fails to include SHTML files in its default list of dangerous files, allowing unrestricted uploads of these files. On servers configured to execute SHTML files, this can lead to remote code execution. **Recommendations** Update Joomla! Core versions 1.0.0 through 5.4.7 to a version newer than 5.4.7. Update Joomla! Core versions 6.0.0 through 6.1.2 to a version newer than 6.1.2.

**Name of the Vulnerable Software and Affected Versions** FileRun versions prior to 2026.2.1 **Description** Authenticated attackers can achieve remote code execution through an OS command injection. The issue occurs when the thumbnail generation system passes filenames wrapped in shell double-quotes directly to the `exec()` function without using `escapeshellarg()` sanitization. This allows filenames containing shell command substitution sequences, such as `$(PAYLOAD).mp4`, to bypass the filename sanitizer and be executed as shell commands when processed by ffmpeg, ImageMagick, vips, or stl-thumb. **Recommendations** Update FileRun to a version later than 2026.2.0.

**Name of the Vulnerable Software and Affected Versions** Sonic 3 A.I.R. versions prior to commit 2492d18 **Description** An unbounded memory allocation issue exists in the `ReceivedPacketCache::enqueuePacket()` function. Unauthenticated remote attackers can crash the server process by sending a specially crafted UDP packet where the `mUniquePacketID` variable is set to the maximum uint32 value. Because the `mUniquePacketID` field is read from the UDP wire-format packet header without bounds checking, the server allocates one CacheItem for every missing packet ID gap. This leads to the exhaustion of available host memory and triggers an uncaught `std::bad alloc` exception, resulting in a call to `std::terminate()`. **Recommendations** Update Sonic 3 A.I.R. to commit 2492d18 or a later version.

**Name of the Vulnerable Software and Affected Versions** Sonic 3 A.I.R. versions prior to commit 2492d18 **Description** A missing source address validation issue exists in the `ConnectionManager` where established connections are resolved using only a two-byte local connection handle. The system fails to verify if the datagram source address matches the registered remote address for the connection. An on-path attacker capable of observing cleartext UDP traffic can inject arbitrary packets into established sessions by forging the two-byte connection identifier. This allows for session termination via `TerminateConnectionPacket`, arbitrary channel message forgery, and forged request responses without the need for IP address spoofing. **Recommendations** Update Sonic 3 A.I.R. to commit 2492d18 or a later version.

**Name of the Vulnerable Software and Affected Versions** Monsta FTP versions prior to 2.14.5 **Description** An unauthenticated server-side request forgery (SSRF) exists due to an incomplete IP blocklist check in the `isBlockedIP()` function, which fails to detect embedded IPv4 addresses within IPv4-mapped IPv6 addresses. An attacker can obtain a CSRF token from the `getSystemVars` endpoint and submit a request to the `fetchRemoteFile` action using a source URL that resolves to an IPv4-mapped address. This allows the server to issue HTTP requests to internal services and write the responses to an attacker-controlled FTP destination, potentially exposing cloud instance metadata credentials. **Recommendations** Update to version 2.14.5 or later. Restrict access to the `fetchRemoteFile` action as a temporary mitigation measure.

**Name of the Vulnerable Software and Affected Versions** NVIDIA Spatial Intelligence Lab's (SIL) GEN3C (affected versions not specified) **Description** The inference API server contains an unauthenticated remote code execution flaw. The endpoints '/request-inference' and '/seed-model' deserialize raw HTTP request bodies using the Python `pickle.loads()` function without requiring authentication or performing input validation. An attacker can send a crafted payload containing a ` reduce ` gadget to the inference API port to execute arbitrary code with the privileges of the inference process. **Recommendations** At the moment, there is no information about a newer version that contains a fix for this vulnerability.

**Name of the Vulnerable Software and Affected Versions** NetBox versions 4.3.5 through 4.5.4 **Description** An issue in the `RenderTemplateMixin.get environment params()` method allows authenticated users with `exporttemplate` or `configtemplate` permissions to execute arbitrary code. By specifying malicious Python callables in the `environment params` field, attackers can bypass Jinja2 SandboxedEnvironment protections. This is achieved by setting the `finalize` parameter to an importable Python callable, such as `subprocess.getoutput`, which is then invoked on every rendered expression outside the sandbox's call interception mechanism, resulting in remote code execution as the NetBox service user. **Recommendations** Update NetBox to a version later than 4.5.4. As a temporary workaround, restrict the `exporttemplate` and `configtemplate` permissions to only trusted users.

**Name of the Vulnerable Software and Affected Versions** OpenCATS versions prior to commit 3002a29 **Description** An unauthenticated PHP code injection issue exists in the installer AJAX endpoint. This allows attackers to execute arbitrary code by injecting PHP statements into the `databaseConnectivity` action parameter. By using a single quote and statement separator, an attacker can break out of the `define()` string context in the config.php file. The injected malicious code persists and executes on every subsequent page load as long as the installation wizard remains incomplete. **Recommendations** Update to commit 3002a29 or a newer version.

**Name of the Vulnerable Software and Affected Versions** LeRobot versions prior to 0.6.0 **Description** An unsafe deserialization issue exists in the async inference pipeline of the policy server and robot client components. The software uses the `pickle.loads()` function to deserialize data received over unauthenticated gRPC channels that lack TLS (Transport Layer Security), a protocol used to encrypt communications. An unauthenticated network-reachable attacker can achieve arbitrary code execution on the server or client by sending a specially crafted pickle payload through the following gRPC calls: - `SendPolicyInstructions` - `SendObservations` - `GetActions` Exploitation of this issue could lead to the compromise of the PolicyServer host and connected robots, theft of sensitive data such as API keys and SSH credentials, lateral movement within the network, denial of service, or threats to physical safety due to the control of robotic systems. **Recommendations** Update LeRobot to version 0.6.0. As a temporary mitigation, restrict network access to the PolicyServer port to prevent unauthenticated remote access.

**Name of the Vulnerable Software and Affected Versions** KTransformers versions prior to 0.5.4 **Description** In the `balance serve` backend mode, the scheduler RPC server binds a ZMQ ROUTER socket to all interfaces without authentication. The server uses the `pickle.loads()` function to deserialize incoming messages without proper validation. This allows an attacker to send a specially crafted pickle payload to the exposed ZMQ socket, leading to arbitrary code execution on the server with the privileges of the ktransformers process. **Recommendations** Update to a version newer than 0.5.3. As a temporary mitigation, restrict access to the ZMQ socket or avoid using the `balance serve` backend mode until the update is applied.

Name of the Vulnerable Software and Affected Versions V2Board versions 1.6.1 through 1.7.4 and Xboard versions through 0.1.9 Description V2Board and Xboard are affected by an issue where authentication tokens are exposed in the HTTP response bodies of the `loginWithMailLink` endpoint when the `login with mail link enable` feature is active. An unauthenticated attacker can send a POST request to the `loginWithMailLink` endpoint with a known email address and receive a full authentication URL in the response. This URL can then be used to exchange the token at the `token2Login` endpoint, granting the attacker a valid bearer token with full account access, including administrative privileges. Recommendations For V2Board versions 1.6.1 through 1.7.4, disable the `login with mail link enable` feature. For Xboard versions through 0.1.9, disable the `login with mail link enable` feature.

Name of the Vulnerable Software and Affected Versions Windmill versions 1.56.0 through 1.614.0 Description Windmill versions 1.56.0 through 1.614.0 have a missing authorization vulnerability. Users with the Operator role can perform prohibited entity creation and modification actions via the backend API. The API does not enforce the Operator restriction on workspace endpoints, allowing Operators to create and update scripts, flows, apps, and raw apps. Operators can also execute scripts via the jobs API, potentially leading to remote code execution within the Windmill deployment. Recommendations Update Windmill to a version later than 1.614.0.

Name of the Vulnerable Software and Affected Versions Windmill CE and EE versions 1.276.0 through 1.603.2 Description Windmill CE and EE versions 1.276.0 through 1.603.2 contain an SQL injection vulnerability in the folder ownership management functionality. Authenticated attackers can inject SQL through the `owner` parameter. This injection allows attackers to read sensitive data, such as the JWT signing secret and administrative user identifiers, forge an administrative token, and then execute arbitrary code via the workflow execution endpoints. Recommendations Windmill CE and EE versions 1.276.0 through 1.603.2 should be updated to a newer version that contains a fix for this vulnerability. As a temporary workaround, consider restricting access to the folder ownership management functionality.

**Name of the Vulnerable Software and Affected Versions** OmniGen2-RL (affected versions not specified) **Description** OmniGen2-RL contains an unauthenticated remote code execution issue in the reward server component. Remote attackers can execute arbitrary commands by sending malicious HTTP POST requests. The root cause is insecure deserialization of request bodies using Python’s pickle module. The vulnerability does not require authentication, simplifying exploitation. The specific HTTP endpoint involved has not been detailed. **Recommendations** At the moment, there is no information about a newer version that contains a fix for this vulnerability.

**Name of the Vulnerable Software and Affected Versions** Keygraph Shannon (affected versions not specified) **Description** Keygraph Shannon contains a hard-coded API key within its router configuration. When the router component is enabled and accessible, network attackers can use this publicly known key to authenticate. An attacker reaching the router port can proxy requests through the Shannon instance, utilizing the victim’s upstream provider API credentials. This can lead to unauthorized API usage and potential exposure of data within proxied requests and responses. The general exploitability of this issue has been mitigated with commit 023cc95. The affected API endpoint is the router configuration. **Recommendations** At the moment, there is no information about a newer version that contains a fix for this vulnerability.

**Name of the Vulnerable Software and Affected Versions** openDCIM versions through 23.04 commit 4467e9c4 **Description** The software contains a SQL injection issue in the `Config::UpdateParameter` function. The `install.php` and `container-install.php` handlers directly incorporate user-provided input into SQL statements using string interpolation, lacking prepared statements or adequate input sanitization. An authenticated user can execute arbitrary SQL statements against the database. The vulnerable parameters are passed to the SQL statements without proper validation. **Recommendations** Versions prior to 23.04 commit 4467e9c4 should be updated.