Base de CVEs

IBM Langflow OSS 1.0.0 through 1.11.1 allows an authenticated attacker to execute arbitrary operating system commands in the server process by saving a flow with a crafted type field value and triggering a build of a wrapper flow that references it. This allowed privilege escalation from "authenticated flow user" to arbitrary OS-level command execution under the server process identity, bypassing the LANGFLOW_ALLOW_CUSTOM_COMPONENTS=false policy control.

SENAITE.CORE is the core framework for the SENAITE laboratory information management system. From 2.0.0 to 2.6.0, the SENAITE.CORE JSON API permits unauthenticated remote code execution through a two-request chain involving missing authorization and unsafe evaluation. The state-changing routes in src/bika/lims/jsonapi/update.py, including update, update_many, remove, doActionFor, doActionFor_many, and getusers, do not enforce the senaite.core: Access JSON API permission before resolving attacker-selected objects. In src/bika/lims/jsonapi/init.py, set_fields_from_request passes raw request values for RecordsField and RecordField instances to eval() before field mutator write-permission checks execute. An anonymous attacker can discover the bika_setup object identifier through @@uuid, send a value such as RejectionReasons to /@@API/update, and execute arbitrary Python in the Zope worker before a later mutation failure rolls back ZODB changes. The same unsafe evaluation pattern is present in src/senaite/core/browser/fields/record.py and src/senaite/core/browser/fields/records.py. Successful exploitation can expose or modify laboratory data, files, and accounts and can disrupt the service.

Improper Neutralization of Directives in Dynamically Evaluated Code ('Eval Injection') in the default lf.query Python protocol in Google langfun versions prior to 0.1.2 allows remote unauthenticated attackers to execute arbitrary Python code in the context of the host application via crafted prompt inputs that cause the model to generate executable Python expressions evaluated without a sandbox.

Xinference is an inference API for running open-source, speech, and multimodal models. In 2.5.0 and earlier, Xinference passes attacker-influenced Llama3 tool-call output to eval() in xinference/model/llm/tool_parsers/llama3_tool_parser.py and xinference/model/llm/utils.py. Requests to /v1/chat/completions with a tools field flow through xinference/api/restful_api.py, xinference/model/llm/transformers/core.py, handle_chat_result_non_streaming(), and _post_process_completion() before extract_tool_calls() or _eval_llama3_chat_arguments() evaluates the model-generated Python expression. An unauthenticated remote attacker can influence that output through a crafted prompt and execute commands in the Xinference server process context. This issue is fixed in version 2.7.0.

In the Neptune connector, a user with access to Neptune through Athena Federated Query could gain access to properties in the Lambda supplying the compute for the connector. To remediate this issue, users should upgrade to aws-athena-query-federation v2026.30.1 or later.

openssl_encrypt versions before 1.4.0 contain a sandbox escape vulnerability in IsolatedPluginExecutor that exposes Python type objects in restricted exec() builtins. Attackers can traverse the Python class hierarchy via __class__.__mro__.__subclasses__() to access system functions and execute arbitrary OS commands.

A remote code execution vulnerability exists in Tenable Security Center's report generation functionality. An authenticated, non-administrative user could exploit this issue by supplying specially crafted input that is later processed unsafely during server-side report rendering, resulting in arbitrary code execution with the privileges of the service account.

Flowise before 3.1.3 contains a sandbox escape vulnerability in the vm2 JavaScript sandbox that allows authenticated users to execute arbitrary code by exploiting moment locale validation bypass. Attackers can craft a fake String object with a match function that bypasses path traversal checks to load and execute malicious JavaScript files stored in the document store outside the sandbox.

Fooocus is an image generating software. In versions 2.5.5 and prior, the Fooocus web UI is vulnerable to remote code execution due to the unsafe use of eval when processing metadata JSON. An attacker with access to the Fooocus web UI may be able to execute arbitrary code on the instance. As of time of publication, no known patched versions are available, but a suggested fix pull request is available.

Adobe Campaign Classic (ACC) is affected by an Improper Neutralization of Directives in Dynamically Evaluated Code ('Eval Injection') vulnerability that could result in arbitrary code execution in the context of the current user. A low-privileged attacker could exploit this vulnerability to execute arbitrary code. Exploitation of this issue does not require user interaction. Scope is changed.

OpenEMR through 8.2.0 contains a remote code execution vulnerability in the document category tree component (library/classes/Tree.class.php) that allows authenticated administrators to execute arbitrary operating system commands by injecting PHP payloads into the categories database table. Attackers can chain arbitrary SQL execution to alter the id column type to VARCHAR and insert a malicious PHP payload, which is then executed via an unsanitized eval() call whenever any page instantiates CategoryTree, including unauthenticated and low-privilege pages, resulting in command execution as the web server user.

SGLang contains an RCE vulnerability when the optional dumper subsystem is enabled, allowing for a sandbox escape when DUMPER_SERVER_PORT is set, enabling code execution on inference requests.

vBulletin 5.x through 5.7.5 and 6.x through 6.2.1 contains an eval injection vulnerability in the vB5_Template_Runtime::runMaths() method within the template runtime that allows unauthenticated remote attackers to execute arbitrary PHP code by supplying crafted input through the pagenav[pagenumber] parameter. Attackers can exploit the insufficiently restrictive regex filter by using phpfuck-style encoding with permitted characters to inject and execute arbitrary PHP code via the unauthenticated ajax/render template route without any authentication.

PraisonAI is a multi-agent teams system. Prior to version 4.6.40, PraisonAI's first-party A2A server example exposes an unauthenticated A2A JSON-RPC endpoint and registers a `calculate(expression)` tool implemented with Python `eval()`. The example also binds to `0.0.0.0`. A remote unauthenticated attacker can send `message/send` to `/a2a`; the request reaches `agent.chat()`, and a real LLM can invoke the registered `calculate` tool. In testing with `gemini/gemini-2.5-flash-lite`, this resulted in arbitrary Python execution in the server process, confirmed by creation of a marker file from an unauthenticated HTTP request. The issue affects deployments following the official A2A example or similar unauthenticated public A2A deployments with unsafe tools. The default unauthenticated A2A surface also exposes task history and task cancellation APIs, increasing confidentiality and integrity impact. Version 4.6.40 patches the issue.