CVE Watch16 de out. de 2026

CVE-2026-54525

Markdown Preview Enhanced is a Markdown preview extension for Visual Studio Code. Prior to 0.8.28, Markdown Preview Enhanced preview-link ha

CVSS

8.8

Alta

EPSS

0.4%

p37

KEV

—

Exploit Today

0-100

Publicado: 14 de out. de 2026 · Última mod.: 16 de out. de 2026 · CWE-78

EPSS · 30d

0.4%EPSS · 30 dias0.4%

2026-10-152026-11-06

Descrição técnica

Markdown Preview Enhanced is a Markdown preview extension for Visual Studio Code. Prior to 0.8.28, Markdown Preview Enhanced preview-link handling passes a decoded, Markdown-controlled link target to a Windows file-opening helper that invokes child_process.exec with the start command. When a Windows user opens an untrusted Markdown file and clicks a crafted external or unsupported-scheme link, shell metacharacters in the decoded target are interpreted by cmd.exe as command separators, allowing an attacker-selected command to run with the privileges of the current VS Code user. The attack requires one normal preview-link click but does not require malicious workspace configuration, script execution to be enabled, or additional files. This issue is fixed in version 0.8.28.

Referências oficiais

CVEs relacionados

CVECVSSEPSSKEVRExploitTítuloMod.

CVE-2026-174089.8 CRÍ

—

———IBM API Connect 10.0.8.0 through 10.0.8.12 and 12.1.0.0 through 12.1.1.2 could allow a remote attacker to execute arbitrary code due to improper neutralization of special elements in an OS command.4h CVE-2026-174029.9 CRÍ

—

———IBM API Connect 10.0.8.0 through 10.0.8.12 and 12.1.0.0 through 12.1.1.2 could allow a remote authenticated attacker to execute arbitrary commands due to OS command injection. User-supplied input is not properly neutralized before it is passed to the operating system, which could allow commands to run on the underlying host.4h CVE-2026-1026106.4 MÉD

—

———fast-agent is a framework for building, running, and evaluating AI agents with Model Context Protocol support. Prior to 0.7.21, fast-agent serve uses HTTP transport bound to 0.0.0.0 on port 8000 without authentication, allowing any network client that can reach the /mcp endpoint to invoke the exposed agent tool. When the server is started with --shell, the agent can call the local execute tool implemented by tools/local_shell_executor.py, so an unauthenticated remote client can cause operating-system commands to run as the server user; without --shell, the client can still consume the operator's model and configured MCP tools. The defaults are defined in cli/commands/serve.py and mcp/server/agent_server.py, and the non-interactive server permission prompt does not prevent the described access. This issue is fixed in version 0.7.21.12h CVE-2026-330508.8 ALT

—

——0GeoNode is a geospatial content management system, a platform for the management and publication of geospatial data. Prior to 5.1.0, a crafted vector file layer name supplied during upload allows a low-privileged remote user to inject shell syntax and execute arbitrary operating-system commands on the GeoNode server. This issue is fixed in version 5.1.0.12h CVE-2026-1233570.0 SEM

—

——0GROBID is machine learning software for extracting information from scholarly documents. Prior to 0.9.1, the Linux non-server PDF conversion path in grobid-core/src/main/java/org/grobid/core/document/DocumentSource.java passes an attacker-influenced PDF filename through DocumentSource.pdfalto, which interpolates the path into a bash -c command string. A single quote in the filename can terminate the intended quoting, and ProcessRunner.run in grobid-core/src/main/java/org/grobid/core/process/ProcessRunner.java starts the resulting command through ProcessBuilder, allowing arbitrary commands to execute with GROBID process privileges. The bundled REST server path is not affected. This issue is fixed in version 0.9.1.14h CVE-2026-1238557.5 ALT

—

——0In JetBrains Toolbox App before 3.6.1 command injection was possible via enterprise configuration URLs4h