Loading vulnerability…
CVE-2026-35020 is an OS command injection vulnerability in Anthropic Claude Code CLI and the Claude Agent SDK. The flaw is described as residing in the command lookup helper and the deep-link terminal launcher, where attacker-controlled input from the TERMINAL environment variable is incorporated into shell command construction and executed via /bin/sh with shell=true. Because shell metacharacters embedded in TERMINAL are interpreted by the shell, a local attacker can inject arbitrary commands. The issue can be reached during normal CLI execution as well as through the deep-link handler path, leading to command execution in the security context of the user running the CLI.
Mallory correlates every CVE against your assets, your vendors, and active adversary campaigns. Know which vulnerabilities matter for you, not just which ones are loud.
What it means. What to do now. Patch path, mitigations, and the assume-compromise checklist.
What an attacker gets, and what they’ve been doing with it.
If you can’t patch tonight, do this now.
Patch, then assume compromise.
No public exploits tracked yet. Mallory keeps watching.
No public exploit code observed for this vulnerability.
Products and vendors Mallory has correlated with this vulnerability. Open in Mallory to drill down to specific CPE configurations and version ranges.
Vendor-confirmed product mapping. Mallory continuously reconciles this list against your asset inventory.
4 sources tracked across advisories, community write-ups, and news. New activity surfaces here as Mallory finds it.
Query your assets running an affected version, and investigate the blast radius.
Every observed campaign linking this CVE to a named adversary.
Malware families riding this exploit, with evidence and IOCs.
YARA, Sigma, Snort, and vendor rules, auto-deployed to your SIEM.
Cross-references every affected SKU, including bundled OEM variants.
Community discussion across Reddit, Mastodon, and other social sources.