CVE-2026-65698

A flaw was found in Void through 1.3.4 in the AI agent file-reading tools (read_file, ls_dir, get_dir_tree, and search_*). Those tools do not confine paths to the open workspace and can accept absolute paths or file:// URIs, including bypassing the approval gate. An attacker who can inject instructions into content the agent processes may read arbitrary host files and exfiltrate sensitive data such as SSH keys or cloud credentials.

Void is vulnerable to path traversal in AI agent file tools that lack workspace confinement. A remote attacker who can get malicious instructions into agent-processed content (user interaction, high attack complexity) may read files outside the workspace via absolute paths or file:// URIs and exfiltrate them through later tool calls. Affects Void through 1.3.4.

The CVSS score(s) listed for this vulnerability do not reflect the associated product's status, and are included for informational purposes to better understand the severity of this vulnerability.