OpenClaw Agents Vulnerable to Phishing and Prompt Injection Attacks
Key Questions
What vulnerabilities have been found in OpenClaw agents?
Two main issues were identified: phishing attacks that can trick agents into leaking credentials and prompt injections that manipulate agent behavior. New vectors include hidden commands in shared contacts and social engineering emails.
What mitigations are recommended for OpenClaw security risks?
Users should update to version 2026.4.23, limit agent autonomy, require human approval for high-risk actions, and apply input sanitization. A new security firewall called Claw Patrol by Deno has also been released.
What does the Claw Crew report say about OpenClaw exposure?
The report shows a 79% prompt injection success rate, 35.4% of OpenClaw instances exposed, and over 824 malicious skills on ClawHub. It includes a practical hardening checklist for solo operators.
Should users consider migrating away from OpenClaw agents?
Related coverage discusses migrating remaining OpenClaw agents to Hermes as one option amid ongoing security concerns. Host-level CLIs are noted as temporary rescue tools only.
Why is immediate attention needed for OpenClaw deployments?
Tens of thousands of setups remain vulnerable to prompt injection with high success rates reported. Safe deployment requires prompt application of the listed mitigations and updates.
Two distinct vulnerabilities: phishing tricks agents into leaking credentials, prompt injections manipulate agent behavior. Attack vectors include hidden commands in shared contacts and social engineering emails. Mitigation requires updating to version 2026.4.23, limiting agent autonomy, human approval for high-risk actions, and input sanitization. Claw Patrol firewall by Deno released. A recent Claw Crew report adds 79% prompt injection success rate, 35.4% exposed instances, 824+ malicious skills on ClawHub, with hardening checklist. Host-level CLIs are temporary rescue tools. AgentGG discovered 5 zero-days, adding urgency.