Decepticon: Role-boundary forgery via ChatML special-token literals in web crawl output composed into LLM context

SummaryDecepticon wraps web crawl results — the output of agent reconnaissance against target services — into LLM messages without neutralizing ChatML special-token literals. Under the BYOK Bring Your Own Key deployment model, users configure their own LLM credentials to any OpenAI-compatible...

PYSEC-2026-4033 Decepticon: Role-boundary forgery via ChatML special-token literals in web crawl output composed into LLM context

Summary Decepticon wraps web crawl results — the output of agent reconnaissance against target services — into LLM messages without neutralizing ChatML special-token literals. Under the BYOK Bring Your Own Key deployment model, users configure their own LLM credentials to any OpenAI-compatible...

PYSEC-2026-4032 Decepticon: Role-boundary forgery via ChatML special-token literals in web crawl output composed into LLM context

Summary Decepticon wraps web crawl results — the output of agent reconnaissance against target services — into LLM messages without neutralizing ChatML special-token literals. Under the BYOK Bring Your Own Key deployment model, users configure their own LLM credentials to any OpenAI-compatible...

Decepticon: Role-boundary forgery via ChatML special-token literals in web crawl output composed into LLM context

SummaryDecepticon wraps web crawl results — the output of agent reconnaissance against target services — into LLM messages without neutralizing ChatML special-token literals. Under the BYOK Bring Your Own Key deployment model, users configure their own LLM credentials to any OpenAI-compatible...

Decepticon: Role-boundary forgery via ChatML special-token literals in web crawl output composed into LLM context

SummaryDecepticon wraps web crawl results — the output of agent reconnaissance against target services — into LLM messages without neutralizing ChatML special-token literals. Under the BYOK Bring Your Own Key deployment model, users configure their own LLM credentials to any OpenAI-compatible...

EUVD-2026-90920

A flaw was found in Foreman. This vulnerability allows an authenticated user with low-level Viewer permissions to cause unauthorized information disclosure by submitting requests to template preview endpoints. By exploiting this issue, the user can access sensitive data, such as host root...

CVE-2026-96659 Foreman: excessive permissions for viewer role on preview

A flaw was found in Foreman. This vulnerability allows an authenticated user with low-level Viewer permissions to cause unauthorized information disclosure by submitting requests to template preview endpoints. By exploiting this issue, the user can access sensitive data, such as host root...

CVE-2026-96659

A flaw in Foreman allows an authenticated user with low-level Viewer permissions to cause unauthorized information disclosure by submitting requests to template preview endpoints. This vulnerability can expose sensitive data, including host root passwords . Additionally, if Safemode protections a...

foreman: Excessive Permissions for Viewer Role on Preview

A flaw was found in Foreman. This vulnerability allows an authenticated user with low-level Viewer permissions to cause unauthorized information disclosure by submitting requests to template preview endpoints. By exploiting this issue, the user can access sensitive data, such as host root...

USN-8857-1 kf6-kcoreaddons vulnerability

It was discovered that KCoreAddons incorrectly handled shell argument quoting in KShell::quoteArgs. The parsing did not adequately handle shell metacharacters, which could lead to a shell escape. An attacker could possibly use this issue to execute arbitrary commands in applications that relied o...

USN-8857-1: KCoreAddons vulnerability

It was discovered that KCoreAddons incorrectly handled shell argument quoting in KShell::quoteArgs. The parsing did not adequately handle shell metacharacters, which could lead to a shell escape. An attacker could possibly use this issue to execute arbitrary commands in applications that relied o...

CVE-2026-76146

An OS command injection vulnerability in Genian SSL PNS allows an attacker who knows only the client access ID, without the password, to execute arbitrary commands remotely...

Rocky Linux 8 [CIQ] Security Update: sg3_utils / sg3_utils-debuginfo / sg3_utils-debugsource / etc (crlsa-2026_56130)

The Rocky Linux 8 host has packages installed that are affected by a vulnerability as referenced in the CIQ crlsa-202656130 advisory. sg3utils: sg3utils: arbitrary command execution via udev property injection in sginq --export CVE-2026-16313 Bug Fixes and Enhancements: sginq output conformance f...

Rocky Linux 8 [CIQ] Security Update: openssh / openssh-askpass / openssh-askpass-debuginfo / etc Multiple Vulnerabilities (crlsa-2026_13383)

The Rocky Linux 8 host has packages installed that are affected by multiple vulnerabilities as referenced in the CIQ crlsa-202613383 advisory. OpenSSH: OpenSSH: Privilege escalation via scp legacy protocol when not preserving file mode CVE-2026-35385 OpenSSH: OpenSSH: Security bypass via...

PT-2026-104822

It was discovered that KCoreAddons incorrectly handled shell argument quoting in KShell::quoteArgs. The parsing did not adequately handle shell metacharacters, which could lead to a shell escape. An attacker could possibly use this issue to execute arbitrary commands in applications that relied o...

CVE-2026-102369: Improper Authentication

Tapo C120 v1 and C200 V5 do not adequately protect login challenge data or sanitize attacker-controlled input processed by the MacTool handler. An unauthenticated attacker on the same local network can replay login challenge data to obtain an administrative session, enable a privileged service th...

PT-2026-105789

Summary Decepticon wraps web crawl results — the output of agent reconnaissance against target services — into LLM messages without neutralizing ChatML special-token literals. Under the BYOK Bring Your Own Key deployment model, users configure their own LLM credentials to any OpenAI-compatible...

PT-2026-105790

Summary Decepticon wraps web crawl results — the output of agent reconnaissance against target services — into LLM messages without neutralizing ChatML special-token literals. Under the BYOK Bring Your Own Key deployment model, users configure their own LLM credentials to any OpenAI-compatible...

PT-2026-105589

Nuclei: Arbitrary Command Execution via DAST Code Signature Bypass in github.com/projectdiscovery/nuclei...

Rocky Linux 10.1 [CIQ] Security Update: vim-X11 / vim-X11-debuginfo / vim-common / vim-data / vim-debuginfo / etc (crlsa-2026_11389)

The Rocky Linux 10.1 host has packages installed that are affected by a vulnerability as referenced in the CIQ crlsa-202611389 advisory. vim: arbitrary command execution via modeline sandbox bypass CVE-2026-34982 Tenable has extracted the preceding description block directly from the CIQ security...