CVE-2026-26322 in OpenClaw
Summary
by MITRE • 02/20/2026
OpenClaw is a personal AI assistant. Prior to OpenClaw version 2026.2.14, the Gateway tool accepted a tool-supplied `gatewayUrl` without sufficient restrictions, which could cause the OpenClaw host to attempt outbound WebSocket connections to user-specified targets. This requires the ability to invoke tools that accept `gatewayUrl` overrides (directly or indirectly). In typical setups this is limited to authenticated operators, trusted automation, or environments where tool calls are exposed to non-operators. In other words, this is not a drive-by issue for arbitrary internet users unless a deployment explicitly allows untrusted users to trigger these tool calls. Some tool call paths allowed `gatewayUrl` overrides to flow into the Gateway WebSocket client without validation or allowlisting. This meant the host could be instructed to attempt connections to non-gateway endpoints (for example, localhost services, private network addresses, or cloud metadata IPs). In the common case, this results in an outbound connection attempt from the OpenClaw host (and corresponding errors/timeouts). In environments where the tool caller can observe the results, this can also be used for limited network reachability probing. If the target speaks WebSocket and is reachable, further interaction may be possible. Starting in version 2026.2.14, tool-supplied `gatewayUrl` overrides are restricted to loopback (on the configured gateway port) or the configured `gateway.remote.url`. Disallowed protocols, credentials, query/hash, and non-root paths are rejected.
If you want to get best quality of vulnerability data, you may have to visit VulDB.
Analysis
by VulDB Data Team • 02/24/2026
The vulnerability described in CVE-2026-26322 affects OpenClaw, a personal AI assistant platform, and represents a significant security flaw in the Gateway tool's handling of user-supplied URLs. This issue stems from insufficient input validation mechanisms that allow malicious actors to manipulate the `gatewayUrl` parameter during tool execution. The vulnerability operates under the principle of insecure input handling and demonstrates how seemingly benign configuration parameters can be exploited to bypass network security controls. The flaw exists in versions prior to 2026.2.14, indicating that it was a persistent issue that required a specific patch to address the underlying security gap.
The technical implementation of this vulnerability relies on the Gateway tool's acceptance of unvalidated `gatewayUrl` parameters without proper sanitization or restriction mechanisms. When tools execute with `gatewayUrl` overrides, the system attempts to establish WebSocket connections to user-specified targets without sufficient validation checks. This design flaw creates a path for attackers to influence outbound network connections from the OpenClaw host system, potentially enabling them to probe internal network services, reach localhost endpoints, or access private network addresses and cloud metadata services. The vulnerability specifically targets the WebSocket client implementation within the Gateway tool, where the URL parsing and connection establishment processes lack proper protocol validation and address restriction controls.
The operational impact of this vulnerability varies significantly based on deployment environment and access controls. In typical secure deployments, the vulnerability requires authenticated operator privileges or trusted automation access to exploit, as the tool invocation paths that accept `gatewayUrl` overrides are restricted to authorized users. However, in environments where untrusted users can trigger tool calls, this vulnerability becomes a serious concern for network reconnaissance and potential lateral movement. The attack surface expands when considering that the system may attempt connections to localhost services, private network addresses, or cloud metadata endpoints, which could reveal internal network topology, service availability, or sensitive configuration information. The vulnerability enables limited network reachability probing capabilities, where successful WebSocket connections to targets can provide additional attack vectors or information gathering opportunities.
The remediation implemented in version 2026.2.14 addresses the core issue through restrictive validation of tool-supplied `gatewayUrl` parameters. The solution enforces strict restrictions that limit URL overrides to loopback addresses on the configured gateway port or the explicitly configured `gateway.remote.url` endpoint. This approach follows the principle of least privilege by preventing arbitrary URL resolution and connection attempts. The patch rejects disallowed protocols, credentials, query parameters, hash fragments, and non-root paths, effectively eliminating the attack vectors that previously enabled exploitation. This mitigation aligns with security best practices outlined in CWE-20, which addresses improper input validation, and follows the ATT&CK technique T1071.004 for application layer protocol tunneling, where the restriction prevents unauthorized network communication patterns. The implementation represents a proper defense-in-depth approach by validating input at the point of use rather than relying on network-level controls to prevent unauthorized connections.