CVE-2026-53829: OpenClaw < 2026.5.18 - Command Truncation in Exec Approval Display

Synopsis

This is a command-truncation vulnerability in OpenClaw, a tool that routes exec commands through an approval workflow. An authenticated attacker can submit a command with a benign-looking prefix followed by a malicious suffix; the approval UI truncates the display so the approver never sees the malicious portion. Once the approver clicks through, the full command, including the hidden suffix, executes with whatever privileges the exec context carries, enabling unauthorized read, write, or denial-of-service operations against the target. A patched-image rebuild at version 2026.5.18 is available on HarborGuard for environments running an affected version.

HarborGuard Coverage

Exploit Conditions

• Network reachabilityRequired

  The attacker must reach the OpenClaw service over the network to submit the crafted exec command.

• AuthenticationRequired

  Any low-privilege account is sufficient; the attacker only needs a valid login to submit exec requests through the approval workflow.

• Victim interactionRequired

  An approver must review and approve the truncated command display before the malicious suffix executes, making this a social-engineering vector that relies on the approver acting on incomplete information.

• Attack complexityDetail

  Exploit conditions are straightforward and reliable; no race conditions or special environmental factors are required beyond crafting an oversized command string.

Blast Radius

• Reads data accessible to the exec context, including secrets, configuration files, and stored application data.
• Writes or modifies files and database rows reachable by the executing process.
• Terminates or disrupts the affected service by injecting destructive command suffixes after approval.
• Enables lateral movement if the exec context carries credentials or tokens for adjacent systems.