Three command injections in Ansible Lightspeed, plus MongoDB RBAC bypass
CVE-2026-44189/44190/44191 all hit CVSS 7.8: opening a malicious project in VS Code gives an attacker shell access. MongoDB also has an 8.1 RBAC bypass (CVE-2026-13059) that lets low-privilege users read and write across tenant boundaries.
Nothing's on fire, but don't scroll past this one. MongoDB has two 8.1s: an RBAC bypass that lets low-privilege users read and write data they shouldn't touch (CVE-2026-13059), and a memory corruption bug in compute mode (CVE-2026-13072). Red Hat also dropped three command injection bugs in the Ansible Lightspeed VS Code extension, all CVSS 7.8, all triggered just by opening a malicious project.
Today's CVEs
Sorted by urgencyCVE-2026-44191
NVDAn attacker can inject shell commands through the Ansible Lightspeed VS Code extension's container and volume mount settings. The injection fires automatically when the Language Server starts up, so just opening a malicious project is enough. If exploited, the attacker gets full code execution on your machine with whatever privileges VS Code runs under.
- Affected estate
- Developers and automation engineers using the Ansible Lightspeed extension in Visual Studio Code
- How to check
- In VS Code, open the Extensions panel and search for 'Ansible Lightspeed.' Check the installed version number against the vendor advisory.
- Included because
- no user interaction beyond opening a project; RCE; developer workstation risk; CVSS 7.8
- Action
- Update the extension to the latest version from the VS Code marketplace. Review any workspace settings for ansible.executionEnvironment.containerOptions and ansible.executionEnvironment.volumeMounts for unexpected values.
- Why it matters
- Opening a malicious project auto-triggers the injection during Language Server startup, giving an attacker RCE with no additional interaction.
Evidence trail
- NVD: View source
CVE-2026-44190
NVDThe Ansible Lightspeed extension doesn't validate the ansible.python.activationScript setting as a proper file path. An attacker who gets you to open a crafted project can inject arbitrary commands through that setting, giving them full control of your machine under VS Code's user privileges.
- Affected estate
- Developers and automation engineers using the Ansible Lightspeed extension in Visual Studio Code
- How to check
- Check the installed extension version in VS Code's Extensions panel. Also search your workspace .vscode/settings.json files for ansible.python.activationScript entries you didn't set yourself.
- Included because
- user interaction required (opening a project); RCE; command injection; CVSS 7.8
- Action
- Update the extension. Remove or correct any suspicious ansible.python.activationScript values in workspace settings.
- Why it matters
- A poisoned project file can execute arbitrary commands the moment you open or run it, with full user privileges.
Evidence trail
- NVD: View source
CVE-2026-44189
NVDA specially crafted playbook filename with shell metacharacters can trick the Ansible Lightspeed extension into running arbitrary code when you execute the playbook. The extension doesn't sanitize the filename before passing it to a shell. This can lead to data theft, file modification, or full system compromise.
- Affected estate
- Developers and automation engineers using the Ansible Lightspeed extension in Visual Studio Code
- How to check
- Check the installed extension version in VS Code's Extensions panel against the vendor advisory.
- Included because
- user interaction required (running a playbook); RCE via filename injection; CVSS 7.8
- Action
- Update the extension. Do not run playbooks from untrusted repositories until the fix is confirmed in place.
- Why it matters
- Running a playbook with a malicious filename gives the attacker code execution under your VS Code user account.
Evidence trail
- NVD: View source
CVE-2026-13072
NVDWhen compute mode is explicitly enabled on a standalone mongod instance, crafted BSON data in an aggregation pipeline can corrupt memory and crash the process or worse. This is a non-default configuration, so most MongoDB deployments aren't exposed. If you've turned on compute mode, though, this is a real concern at CVSS 8.1.
- Affected estate
- MongoDB operators running standalone mongod instances with compute mode explicitly enabled
- How to check
- Review your mongod startup flags or configuration file for compute mode enablement. Run db.adminCommand({getCmdLineOpts: 1}) to inspect active options.
- Included because
- non-default config required; memory corruption; CVSS 8.1; no exploit in the wild
- Action
- Upgrade MongoDB to the patched version. As a workaround, disable compute mode if it's not strictly needed.
- Why it matters
- Memory corruption from crafted BSON can crash your database process or potentially allow further exploitation.
Evidence trail
- NVD: View source
CVE-2026-13059
NVDA low-privilege authenticated user can bypass role-based access controls and read or write data they shouldn't have access to. The bug hits find, update, delete, and aggregate commands when apiStrict mode isn't enabled. If you rely on MongoDB's RBAC to enforce data boundaries between users or tenants, this one breaks that trust boundary.
- Affected estate
- MongoDB operators relying on role-based query-level access controls in non-apiStrict configurations
- How to check
- Check whether apiStrict is enabled in your connection options or mongod/mongos configuration. If it's not, and you use document-level RBAC, you're exposed.
- Included because
- authenticated but low-privilege; RBAC bypass for reads and writes; CVSS 8.1; common product
- Action
- Upgrade MongoDB to the patched version. If patching is delayed, enable apiStrict mode to block the vulnerable command paths.
- Why it matters
- Any authenticated user, even one with minimal privileges, can read and modify data outside their authorized scope, breaking your access control model.
Evidence trail
- NVD: View source
One email, every Wednesday morning.
SubscribeFrom the field notes
From this beat
Read the rest of the field notes โ