Spring AI's support for Anthropic's Skills API uses LLM‑influenced filenames without sanitization in Path.resolve before the file is written to disk, which allows files to be written outside the intended target directory, including restricted directories. This directory traversal flaw permits an attacker to create or overwrite files beyond the intended scope, potentially affecting the integrity of the system.

Affected: Spring AI, versions 1.1.0 through 1.1.x. The issue is present in all releases of Spring AI that include the Anthropic Skills API integration within that range.

The flaw carries a CVSS score of 6.5, indicating medium severity. EPSS score is not available and the vulnerability is not listed in the CISA KEV catalog. The likely attack surface is remote; an attacker who can invoke the Anthropic Skills API endpoint could supply a crafted filename that causes the vulnerable Path.resolve logic to write the file outside the intended directory, allowing unauthorized creation or modification of files in locations beyond the intended scope.