The OpenClaw Agent Platform v2026.2.6 contains a remote code execution vulnerability caused by a request‑side prompt injection flaw. An attacker can manipulate input sent to the platform, causing the executed server‑side code to run arbitrary commands. This flaw maps to CWE‑94, representing code injection. The consequence is full compromise of the affected host, allowing the attacker to gain unrestricted access, install malware, and exfiltrate data.

Affected systems are those running OpenClaw Agent Platform version 2026.2.6. The product is identified by the CPE cpe:2.3:a:openclaw:openclaw:*:*:*:*:*:node.js:*:*. No other vendors or products are listed or affected.

The CVSS base score is 9.8, indicating critical severity. The EPSS score is below 1%, suggesting low current exploitation probability, but the flaw remains a top‑priority vulnerability. The issue is not listed in the CISA KEV catalog. Attackers would need network access to the vulnerable endpoint that accepts user input for the prompt injection; exploitation would be remote and requires no authentication.