vulnerability

CVE-2026-90898 lets unauthenticated attackers execute arbitrary commands on Bifrost AI gateway servers via a single HTTP request; fixed in transports/v2.1.0.

Security firm Air Security found a zero-click flaw letting a plugin repository owner bypass commit-hash pinning in four AI coding agents.

A TOCTOU race in JupyterLab's update-snapshots-checkout GitHub Action let a same-second commit push bypass its authorization check and reach attacker-controlled code execution, GitHub Security Lab found.

CISA added CVE-2026-82329, a 9.8-severity Artifactory authentication bypass, to its exploited-vulnerabilities catalog days after JFrog patched it.

Manifold Security found eight flaws across seven AI coding agents that let a repository's git config silently execute code the moment it is opened.

CISA added Gitea's CVE-2026-60004 remote code execution flaw to its KEV catalog after BleepingComputer reported attackers deploying cryptomining malware on unpatched servers.

A crafted marimo notebook could run an attacker's command as a local subprocess the moment it was opened in edit mode, via a fake MCP server entry in its configuration.

A critical Jinja2 template-injection bug let attacker-controlled repository files execute code in the Serena coding agent, bypassing its own trust gate.

A CVSS 9.8 Gitea flaw let unauthenticated users read server files via Org-mode markup; XBOW Security's autonomous system found it.

GitLab's 19.2.1, 19.1.3, and 19.0.5 patch release fixes 13 vulnerabilities, including three high-severity flaws and a prompt-injection bug in Duo Code Review.

CISA gave federal agencies until August 6 to patch CVE-2026-18577, an N-central auth bypass exploited in the wild since July 31 that grants attackers admin access to the RMM console.

CVE-2026-66066 lets attackers upload a crafted image to steal a Rails app's secret_key_base and escalate to remote code execution.