Signal is active with 1 mentions in latest observed window
Recommended action window: Monitor and triage in normal cycle
NVD description
Claude Code is an agentic coding tool. In versions prior to 2.1.75 on Windows, Claude Code loaded the system-wide default configuration from C:\ProgramData\ClaudeCode\managed-settings.json without validating directory ownership or access permissions. Because the ProgramData directory is writable by non-administrative users by default and the ClaudeCode subdirectory was not pre-created or access-restricted, a low-privileged local user could create this directory and place a malicious configuration file that would be automatically loaded for any user launching Claude Code on the same machine. Exploiting this would have required a shared multi-user Windows system and a victim user to launch Claude Code after the malicious configuration was placed. This issue has been fixed on version 2.1.75.
Priority
LOW
Exploitation
NONE
PoC
NONE
Patch
NONE
Momentum
STABLE
If you run products in this scope, you should treat this CVE as relevant to your environment.
1 version affected across 2 products
| Date | Total | Labels |
|---|
| 2026-04-18 | 2 | General2 |
| 2026-06-18 | 2 | General2 |
| 2026-09-12 | 1 | Disclosure1 |
If you received this Anthropic email, you were probably affected by a vulnerability we found. CVE-2026-35603 👇 https://t.co/otQKOtkD0w
Post summary
The tweet notes that recipients of a specific Anthropic email were likely affected by CVE-2026-35603, but offers no technical or mitigation details.
Last night I read CVE-2026-35603. An AI coding tool trusts a config file in a folder anyone on a shared machine can write to, so every user on that box runs the attacker's code. I went and looked at where my agent configs load from, and who can write there.
Full research: https://cymulate.com/blog/cve-2026-35603-ai-coding-tools-privilege-escalation/
Post summary
The provided text contains only a link to a research blog post about CVE‑2026‑35603 with no additional information on PoC, exploit, or mitigation details.
An agent, fully inside its sandbox, writes a file the boundary was never designed to police - a .claude hook, a .vscode task config, a modified venv. Something outside the sandbox reads and executes it later. CVE-2026-35603 and CVE-2026-48124 came out of that wave.
Post summary
The text announces the appearance of CVE‑2026‑35603 and CVE‑2026‑48124 linked to sandbox escape via unseen file types but offers no further technical or practical information.
CVE-2026-35603 Claude Code is an agentic coding tool. In versions prior to 2.1.75 on Windows, Claude Code loaded the system-wide default configuration from C:\ProgramData\ClaudeCode… https://www.cve.org/CVERecord?id=CVE-2026-35603
Post summary
The text briefly mentions a CVE and references a version and a configuration path, but provides no actionable details about exploitation, patches, or technical specifics.
🟠 Claude Code, Insecure System-Wide Configuration Loading, #CVE-2026-35603 (Moderate) https://dailycve.com/claude-code-insecure-system-wide-configuration-loading-cve-2026-35603-moderate/
Post summary
The provided text only gives a headline and a link, with minimal technical details about CVE‑2026‑35603 and no evidence of exploitation, PoC, or patch information.
2 of 2 entries
| Part | Vendor | Product | Version | Target SW | Target HW |
|---|---|---|---|---|---|
| App | anthropic | claude_code | - | node.js | - |
| OS | microsoft | windows | - | - | - |