Israeli cybersecurity company Adversa AI has launched SecureClaw, an open-source security platform designed to protect OpenClaw AI agents through automated auditing, hardening, and runtime protection.
What SecureClaw Does
The platform addresses security vulnerabilities in OpenClaw deployments by providing defense-in-depth protection against prompt injection, credential exposure, supply-chain threats, and agent workflow manipulation.
Key Features
- 55 automated audit and hardening checks covering the full OWASP Agentic Security Initiative Top 10.
- MITRE ATLAS mapping — formal mapping to MITRE ATLAS threat categories for enterprise compliance.
- Two-layer architecture — a code-level security plugin and a behavioral security skill work together.
- Protection scope — guards against credential theft, prompt manipulation, and agent takeover attempts.
Why It Matters
"Security for OpenClaw cannot be an afterthought," said Alex Polyakov, Founder and CTO of Adversa AI. "We need to act immediately to secure this innovation."
SecureClaw is the only OpenClaw security solution to formally map protections across all OWASP Agentic Security Initiative categories. With OpenClaw's rapid adoption — over 200,000 GitHub stars — securing deployments has become critical, especially for businesses handling sensitive data.
The plugin is available as open source and can be installed on any OpenClaw deployment, including those managed through OpenClaw Launch.