Loading story…
A critical vulnerability tracked as CVE-2026-73678 allows unauthenticated remote code execution in MindsDB Minds Platform v26.1.0 and earlier. The flaw stems from the exposed POST /api/v1/responses/ endpoint, which can route attacker-controlled prompts to the Anton agent's scratchpad tool, where Python source is executed through exec() without sandboxing. Attackers can chain this with the unauthenticated PUT /api/v1/settings/ endpoint to set their own LLM API key and then run arbitrary Python and operating system commands as the application process or desktop user.
The issue is classified as CWE-94 and was assigned the maximum severity with CVSS v3.1 10.0 and CVSS v4.0 10.0 in one disclosure, reflecting network-based exploitation with no required privileges or user interaction and full impact on confidentiality, integrity, and availability. Successful compromise could expose SSH keys, stored credentials, and environment secrets, and enable complete system takeover in the application's context. At disclosure, no vendor patch and no confirmed in-the-wild exploitation were reported; recommended mitigations included restricting external API access, placing the service behind an authenticating reverse proxy, disabling or firewalling the vulnerable endpoints, tightening CORS, and isolating the platform runtime.
Track how attackers are adapting to this technology.
3 events from the most recent confirmed update back to the earliest known activity.
The CVE entry for CVE-2026-73678 was received by disclosure@vulncheck.com. The vulnerability concerns an unauthenticated remote code execution flaw in MindsDB Minds Platform.
At the time of disclosure, no vendor patch had been published for CVE-2026-73678. Reporting also stated that no active exploitation was known at that time.
CVE-2026-73678 was disclosed as a critical unauthenticated remote code execution vulnerability affecting MindsDB Minds Platform 26.1.0 and earlier. The flaw allows attackers to set an LLM API key via an unauthenticated settings endpoint and then trigger arbitrary Python and OS command execution through the Anton agent's scratchpad exec() path.
Vulnerabilities, threat actors, malware, products, organizations, and breaches Mallory has linked to this story.
Follow how adversaries are adapting to this technology, and where it touches your stack today.
2 references tracked. Mallory keeps watching after this page renders.
Map indicators from this story to your assets and identify affected systems in minutes.
Every observed campaign, victim, and pivot linked to actors named in this story.
Malware, exploits, and IOCs connected to the activity described here.
YARA, Sigma, and Snort rules deployed to your SIEM as soon as they’re published.
Get matching new stories delivered to your team as they break — not the next morning.
Ask questions about this story and take action on the answers.
Other stories Mallory is tracking that overlap on entities, sources, or topic labels.
Two high-severity vulnerabilities were disclosed affecting exposed application and device management surfaces, including a flaw that can give attackers **root access** and another that enables **arbitrary file write** through path traversal. **CVE-2026-3587** describes an unauthenticated remote attack path in a hidden CLI function that lets an attacker escape a restricted prompt and gain root access to the underlying Linux operating system, potentially leading to full device compromise. The issue was mapped to `CWE-912` and assigned a `CVSS v3.1` score vector of `AV:N/AC:L/PR:N/UI:N/S:C/C:H/I:H/A:H`, with CERT VDE publishing advisory `VDE-2026-020`. A separate vulnerability, **CVE-2026-5027**, affects Langflow's `POST /api/v2/files` endpoint, where improper sanitization of the multipart `filename` parameter allows path traversal using `../` sequences. An authenticated attacker can exploit the bug to write files to arbitrary filesystem locations, creating a route to compromise confidentiality, integrity, and availability. The flaw was classified as `CWE-22`, carries the `CVSS v3.1` vector `AV:N/AC:L/PR:L/UI:N/S:U/C:H/I:H/A:H`, and is referenced in Tenable advisory `TRA-2026-26`.
Jun 29, 2026
Several server and developer administration tools have disclosed severe code-execution vulnerabilities affecting exposed management functionality. **Termix** reported an authenticated OS command injection in `GET /ssh/file_manager/ssh/resolvePath`, where double-quoted user input allowed shell substitution via ``$(...)`` and backticks, enabling arbitrary commands on the connected remote host under the active SSH account. A separate **Termix** issue, tracked as `CVE-2026-45746`, affected versions prior to **2.3.2** and allowed attackers to tamper with a client-controlled `sessionId` to access other users' active File Manager SSH sessions, potentially leading to unauthorized file operations and remote code execution on another user's VPS. Other disclosures showed similar high-impact flaws in JavaScript-based backends. **DbGate** fixed an RCE in **v7.1.9** after researchers found that `POST /runners/start` and `POST /runners/load-reader` concatenated attacker-controlled values into generated JavaScript executed by a forked Node.js process; in anonymous-auth deployments, the issue could permit **unauthenticated RCE**. **BrowserStack Runner** through **0.9.5** was assigned `CVE-2026-49143` for an unauthenticated RCE in the `/_log` handler, where crafted JSON passed through `vm.runInNewContext()` and `eval()` allowed attackers to escape the sandbox and invoke `child_process` on reachable runner instances.
Jun 29, 2026
BMC Control-M deployments are affected by two newly disclosed vulnerabilities that expose `Control-M/Server` and related components to serious remote attacks. **CVE-2026-10539** is a critical unauthenticated command injection flaw in Control-M/Server communication commands that can let a remote attacker execute unauthorized commands and potentially fully compromise a server. The issue affects `Control-M/Server` versions `9.0.20.x` through `9.0.21.200`, and possibly earlier unsupported releases, and is rated `CVSS 4.0 9.5` / `CVSS 3.1 9.0`. A second issue, **CVE-2026-10538**, is a high-severity improper deserialization vulnerability in out-of-support `Control-M/Server` and `Control-M/Enterprise Manager` `9.0.20.x` components, particularly in messaging consumer functionality that processes user-controlled serialized data. An authenticated attacker could remotely trigger unintended server-side behavior through crafted serialized content; the flaw is rated `CVSS 4.0 8.9` / `CVSS 3.1 8.0`. Organizations running affected Control-M versions have been advised to upgrade to supported releases, apply vendor patches, harden deserialization controls, restrict access to exposed services, and monitor for suspicious command execution or messaging activity.
Jul 3, 2026