EXECUTIVE SUMMARY:
CVE-2026-41686 with a CVSS score of 5.5 is a vulnerability in the npm/@anthropic-ai/sdk package, specifically affecting versions greater than or equal to 0.79.0 and less than 0.91.1, with the patched version being 0.91.1. The vulnerability arises from the `BetaLocalFilesystemMemoryTool` in the Anthropic TypeScript SDK, which creates memory files and directories using Node.js default modes, resulting in world-readable files and directories on systems with a standard umask or world-writable files in environments with a permissive umask, such as many Docker base images. A local attacker on a shared host could exploit this vulnerability by reading persisted agent state, while in containerized deployments, an attacker could modify memory files to influence subsequent model behavior, thereby gaining the capability to access sensitive data or manipulate the model's behavior. If exploited, this vulnerability could have significant business impacts, including data confidentiality breaches and model manipulation, which could lead to financial losses, reputational damage, and compromised business operations.[/subscribe_to_unlock_form]
EXECUTIVE SUMMARY:
CVE-2026-41686 with a CVSS score of 5.5 is a vulnerability in the npm/@anthropic-ai/sdk package, specifically affecting versions greater than or equal to 0.79.0 and less than 0.91.1, with the patched version being 0.91.1. The vulnerability arises from the `BetaLocalFilesystemMemoryTool` in the Anthropic TypeScript SDK, which creates memory files and directories using Node.js default modes, resulting in world-readable files and directories on systems with a standard umask or world-writable files in environments with a permissive umask, such as many Docker base images. A local attacker on a shared host could exploit this vulnerability by reading persisted agent state, while in containerized deployments, an attacker could modify memory files to influence subsequent model behavior, thereby gaining the capability to access sensitive data or manipulate the model's behavior. If exploited, this vulnerability could have significant business impacts, including data confidentiality breaches and model manipulation, which could lead to financial losses, reputational damage, and compromised business operations.[emaillocker id="1283"]
RECOMMENDATION:
REFERENCES:
The following reports contain further technical details:
https://github.com/advisories/GHSA-p7fg-763f-g4gf