Broken IAM authorization in Floci lets attackers reach Security Hub and other security services without authorization
An attacker can bypass authentication with no policy permission because Floci's IAM enforcement does not cover Security Hub, Macie, and GuardDuty. `IamActionRegistry` lacks the corresponding route rules, and the models are not correctly loaded into `rest-routes.json`, so `resolveRoute` returns an empty action and allows every request by default. An attacker who can reach the service over the network can send malicious HTTP requests to perform sensitive operations such as enabling services or listing administrator accounts. A system under the attacker's control faces serious damage to data confidentiality and integrity, and the security boundary is completely lost.
Component
Floci is a localized cloud-service simulation platform compatible with the AWS API, built to give development and test environments a seamless cloud-native experience. Its core architecture simulates multiple AWS services through RESTful interfaces, including IAM, Security Hub, Macie, and GuardDuty, and supports fine-grained identity and access-management policy enforcement. In the system context it plays a key infrastructure role: it handles resource scheduling and security-audit requests in a multi-account setup, keeps local behavior consistent with the cloud, and thereby improves the organization's cloud-application deployment flow and compliance-check efficiency.
Complete the IAM action registry and include every related service model in the route-resolution configuration. Also update the resource builder so it recognizes the full set of service resource types, and regenerate the authorized-action list, so that every API endpoint goes through strict policy checks and the unauthorized-access risk from missing configuration is removed.
CSSA Exclusive Early Warning
10.0 Critical
A configuration flaw in DesktopCommanderMCP disables command filtering and creates a remote-code-execution risk
The flaw comes from a faulty fallback in configuration-validation logic. When an attacker uses the set_config_value endpoint to set the blockedCommands parameter to an empty array, validateCommand fails to trigger the default block list because of a logic error, and every command-line filter is bypassed. The attacker can send malicious requests remotely over the network and, without authentication, run arbitrary high-risk system commands, fully compromising the host and damaging confidentiality, integrity, and availability.
Component
DesktopCommanderMCP is a desktop automation-agent component based on the Model Context Protocol. Its core architecture is meant to bridge large language models and the local operating system through a standardized interface, providing cross-platform features such as file management, process monitoring, and execution of system commands, and aiming to improve human-computer collaboration and automated resource scheduling inside a security sandbox.
Restructure the configuration-validation logic so that when blockedCommands is empty or null it is forced back to the default safety blacklist. Use an explicit non-empty confirmation to prevent misconfiguration, and set a default deny-all policy for critical security parameters such as allowedDirectories, eliminating the fail-open risk at the root.
CSSA Exclusive Early Warning
10.0 Critical
An authentication flaw in the sglang gateway MCP component breaks multi-tenant isolation and allows unauthorized access
The flaw is in `ensure_request_mcp_client`. When it builds an MCP client, it uses only the server URL as the cache key and ignores differences in the Authorization token, so different tenants or anonymous requests reuse the same already-authenticated session. An attacker can craft a request that points to the same MCP URL and, using the credentials of a legitimate user who connected first, run arbitrary tool operations, steal other tenants' sensitive data, or perform unauthorized actions, causing serious damage to data confidentiality and integrity.
Component
sglang is a high-efficiency serving framework for large-language-model inference. Its core gateway component routes API requests and integrates external tools. The architecture uses a process-level shared MCP manager to optimize resource use, and it is designed to let multiple tenants call external MCP servers concurrently through a unified interface, extending functionality and decoupling business logic.
To fix this authentication-bypass flaw, restructure session management so that credentials are part of the unique key in the connection pool, and each distinct credential has its own session instance. Enforce strict tenant-context isolation so the global tool inventory is not shared across privilege boundaries, and add request-level dynamic configuration checks so a static cache cannot leak authentication state.
CVE Intelligence
2
CVE-2026-78287
CVSS
10.0 Critical
2026-11-11
Code injection in Staggs STAGGS leads to remote code inclusion
In Staggs 3.3.0 and earlier, a core module does not properly control generated code (CWE-94), a code-injection flaw. An attacker can craft malicious input that triggers remote code inclusion and directly threatens system integrity and availability. When Staggs handles certain requests, it does not correctly filter or escape user input, so malicious code is interpreted and executed. Every Staggs user who has not updated past 3.3.0 is affected. The attacker only needs to send a crafted request over the network, with no user interaction, and can operate the system remotely.
Component
Staggs is an open-source software component widely used in various web-application development, providing basic functional support.
Risks
From ordinary user to administrator: If the victim runs the application with administrator privileges, the attacker can obtain the same privileges
Complete system control: The attacker can run arbitrary code on the victim system and, depending on user privileges, install programs, view, change, or delete data, or create new accounts with full privileges
Attack without user interaction: Through a remote network request (T1190), the attacker can trigger the flaw with no user interaction and achieve remote code execution
Enable strict input validation in the runtime environment
Block code-inclusion requests from untrusted sources
CVE-2026-118965
CVSS
9.8 Critical
2026-11-11
Path traversal in Checkmk host-label synchronization leads to remote code execution
In Checkmk versions <2.5.0p19, <2.4.0p40, and 2.3.0 (EOL), host-label synchronization does not adequately restrict path names, a path-traversal flaw (CWE-22). In a distributed monitoring environment, an attacker who controls a remote site can craft a malicious request and write arbitrary files on the central site, then execute code as the central-site user. Every Checkmk user who has not updated to a fixed version is affected, especially environments that deploy a distributed monitoring architecture. The attacker needs no additional user interaction. Control of a remote site is enough to trigger the flaw and achieve remote code execution.
Component
Checkmk is an enterprise-grade open-source monitoring solution based on Nagios. It supports a distributed monitoring architecture and is widely used for performance monitoring and alerting of IT infrastructure.
Risks
From remote site to central site: If the attacker controls a remote site in a distributed environment, they can use this flaw to break isolation and run arbitrary code on the central site
Complete system control: The attacker can run arbitrary commands on the central-site server and, depending on the central-site user's privileges, install programs, view, change, or delete data, or create new accounts with full privileges
Lateral movement in a distributed environment: By using host-label synchronization, the attacker can move laterally from an edge node toward the core node in a distributed monitoring network
Upgrade Checkmk immediately to 2.5.0p19, 2.4.0p40, or later
Apply a strict network access-control list (ACL) to communication between the central site and remote sites
Monitor the central-site filesystem for abnormal write behavior, especially file creation in unexpected directories
Package Poisoning
5
Package Poisoning
npm2026-11-12
chai-as-spy@>= 0
flagged as malicious
This version was found communicating with a malicious domain and executing malicious commands. Audit your dependencies and pin safe versions immediately.
MD5
98a848117a5b73d3cdc05138fb859e47
Package Poisoning
npm2026-11-12
express-helmet@>= 0
flagged as malicious
This version was found communicating with a malicious domain and executing malicious commands. Audit your dependencies and pin safe versions immediately.
MD5
6862f14a1df866367751589bf86e0c64
Package Poisoning
npm2026-11-12
gloglet@>= 0
flagged as malicious
This version was found communicating with a malicious domain and executing malicious commands. Audit your dependencies and pin safe versions immediately.
MD5
9a374c2679180a843d4c4251096dc569
Package Poisoning
npm2026-11-12
shraps@>= 0
flagged as malicious
This version was found communicating with a malicious domain and executing malicious commands. Audit your dependencies and pin safe versions immediately.
MD5
a7f855747a5d26a8085c85566ef17cfb
Package Poisoning
npm2026-11-12
verboset@>= 0
flagged as malicious
This version was found communicating with a malicious domain and executing malicious commands. Audit your dependencies and pin safe versions immediately.
