OpenClaw versions earlier than 2026.5.7 contain a hostname validation flaw in the retry endpoint. The validation logic accepts a hostname that shares a prefix with a trusted host, instead of requiring an exact match. An attacker can craft a hostname with a matching prefix and cause the application to send authentication or session data to an untrusted endpoint. The flaw is a classic example of improper input validation (CWE-1023). As a result, credential information may be exposed to an attacker under the guise of a legitimate service.

The vulnerability affects the OpenClaw OpenClaw product implemented in a node.js environment. All releases before 2026.5.7 are impacted; the bug is reflected in the CPE entry for OpenClaw.

The CVSS score of 6 denotes moderate severity. No EPSS score is available, and the vulnerability is not listed in the CISA KEV catalog. Attackers do not need privileged access; by submitting a crafted request that uses a hostname prefix resembling a trusted host, they can trigger the retry logic and cause authentication material to be transmitted to an attacker-controlled host. If successful, credential theft is possible without further interaction.