the critical ledger · updated hourly

Critical vulnerability intelligence.

New critical security flaws land every day. Vulnary tells you what each one actually is, who it hits, how serious it really is, and the exact steps to shut it down.

Browse the feed ↓Am I affected? →

scroll

2,488critical tracked/74new today/39known-exploited/1exploit available/2,488critical tracked/74new today/39known-exploited/1exploit available/2,488critical tracked/74new today/39known-exploited/1exploit available/2,488critical tracked/74new today/39known-exploited/1exploit available/

The feed (latest)

2,488 entries

Your stack

Add the software you run to see what affects you.

view

source

fromto

127

CVE-2026-79724

IBM Langflow OSS versions 1.0.0 through 1.11.5 contain a critical flaw that lets attackers run any operating‑system commands on the host. The vulnerability is caused by improper handling of special characters in command strings.

Critical· 9.8<1% likely

published 3d agodeletes in 43d

128

CVE-2026-78573

This vulnerability (CVE-2026-78573) affects IBM ContextForge MCP Gateway versions 1.0.0 through 1.0.7, allowing remote attackers to gain administrative access by exploiting default credentials. It is classified as critical (CVSS 9.8) and stems from improper authentication (CWE-1392).

Critical· 9.8<1% likely

published 3d agodeletes in 43d

129

CVE-2026-45764

Suricata versions before 7.0.16 and 8.0.5 are vulnerable to a type‑confusion flaw when processing HTTP/2 traffic, which can be triggered by crafted packets to crash the engine and cause a denial‑of‑service. The issue is fixed in Suricata 7.0.16 and 8.0.5. As a temporary workaround, administrators can disable HTTP/2 parsing if it is not needed.

Critical· 9.1<1% likelyfix available

published 3d agodeletes in 28d

130

CVE-2026-19646

A flaw in IBM Common Licensing Agent and ART versions 9.0 allows a remote attacker to redirect users to any domain by manipulating the HTTP Host header. The vulnerability is exploitable over the network and does not require authentication or user interaction. It could be used to trick users into visiting malicious sites.

Critical· 9.1<1% likely

published 3d agodeletes in 43d

131

CVE-2026-89094

Forgejo versions older than 16.0.4 can run arbitrary code when a user creates a specially crafted template repository. The flaw is due to improper handling of template files in the .forgejo/template directory.

Critical· 9.9<1% likelyfix available

published 3d agodeletes in 28d

132

CVE-2026-85025

IBM Langflow OSS versions 1.0.0 through 1.11.5 contain a critical flaw that lets anyone run arbitrary code and tamper with chat sessions via publicly shared MCP project endpoints.

Critical· 9.8<1% likely

published 3d agodeletes in 43d

133

CVE-2026-75940

A critical flaw in the Lenovo Health Android app, sold only in China, lets attackers read private health data. The weakness is in how the app handles authentication, allowing unauthorized access. Users should be aware that their personal health information could be exposed.

Critical· 9.3<1% likely

published 3d agodeletes in 43d

134

CVE-2026-89086

The OCaml jose library before version 0.11.0 incorrectly validates RSA signatures, accepting any signature that can be decoded with PKCS #1. This flaw allows attackers to forge signatures without needing the public key, potentially leading to unauthorized actions.

Critical· 9.1<1% likelyfix available

published 3d agodeletes in 28d

135

CVE-2026-88062

OmniRoute, an open‑source AI gateway, has a critical flaw that lets attackers run arbitrary code on the server. The vulnerability exists in versions 3.8.49 and earlier, where the POST /api/acp/agents endpoint accepts attacker‑controlled commands. Because the server executes these commands without proper validation, a remote user can run any code on the container.

Critical· 9.5<1% likely

published 3d agodeletes in 43d

136

CVE-2026-89043

The vulnerability is an XML signature wrapping flaw in passport-saml-encrypted 0.1.13 and earlier. It lets an attacker insert a forged SAML assertion that is accepted as the authenticated identity while the original signed assertion remains valid. This can allow an attacker to impersonate users without needing to break the signature.

Critical· 9.1<1% likely

published 3d agodeletes in 42d

137

CVE-2026-89042

The passport-saml-encrypted library (up to version 0.1.13) has a flaw that lets attackers bypass SAML signature checks. By sending unsigned SAML responses, an attacker can trick the system into authenticating them as any user, gaining access to protected profiles.

Critical· 9.3<1% likely

published 3d agodeletes in 42d

138

CVE-2026-88044

A flaw in rclone’s serve/start RC interface caused the program to ignore per-server authentication proxy settings for FTP and S3. As a result, FTP connections defaulted to anonymous access and S3 connections served a fixed filesystem instead of the intended backend. The bug was fixed in rclone version 1.75.1.

Critical· 9.1<1% likelyfix available

published 3d agodeletes in 27d

139

CVE-2026-68488

A race condition in Plesk allows a local attacker to exploit insecure symlink handling, giving them root privileges. The flaw lets the attacker take ownership of any file or directory. This can be used to fully compromise the server.

Critical· 9.9<1% likely

published 3d agodeletes in 42d

140

CVE-2026-68487

A path traversal flaw in Plesk's Backup Manager lets an authenticated customer write arbitrary files as root. This could allow attackers to modify system files or install malware. The flaw is critical with a CVSS score of 9.9.

Critical· 9.9<1% likely

published 3d agodeletes in 42d

141

CVE-2026-65639

This vulnerability allows a remote attacker to run arbitrary commands as root on a system running ConfigServer Security & Firewall. The flaw is in the advanced-rule parser, which fails to validate data from an allow/deny feed. An attacker who can control that feed can execute any command with root privileges.

Critical· 9.52% likelyfix available

published 3d agodeletes in 27d

142

CVE-2026-65638

A flaw in ConfigServer Security & Firewall lets anyone on the network run commands as the CSF service account. The issue is caused by improper URL escaping that allows shell command injection. Updating to the latest version fixes the problem.

Critical· 9.23% likelyfix available

published 3d agodeletes in 27d

143

CVE-2026-52098

A critical flaw in Flowise 3.1.2 lets attackers run any code on the server by hitting a specific API endpoint. The vulnerability can be triggered over the network without authentication or user interaction. It poses a severe risk to any system running that version.

Critical· 9.8<1% likely

published 3d agodeletes in 42d

144

CVE-2026-88899

A critical vulnerability (CVE-2026-88899) allows remote attackers to supply arbitrary directory paths in the x-opencode-directory request header to the /api/opencode endpoint. This flaw lets attackers perform file operations outside the intended project root on the host system. The issue exists in versions before 0.31.0 of the affected software.

Critical· 9.3<1% likelyfix available

published 3d agodeletes in 27d