Security insights for OpenClaw.

v1.5 adds built-in audit integration, credential flow mapping, SSRF detection, sandbox scoring, accept risk, capability drift, network policy generation, and tamper-evident audit trail.

The built-in audit has 78 config checks. It doesn’t scan skills against 1,184+ malicious IOCs, verify identity files, audit MCP servers, or detect persistence. Here’s what covers the rest.

7% of ClawHub skills expose credentials through the LLM context window. Plaintext config is just level 0. Here’s the full hierarchy of API key protection.

Meta, Kakao, Naver, and Karrot Market banned OpenClaw from corporate networks. Google restricted paying subscribers who used it. Here’s why — and what it means if your employees are already running it.

China’s MIIT issued a formal security alert about OpenClaw — while Chinese cloud providers were actively selling hosting packages for it. The full story, what they found, and what it means for global deployments.

Belgium, China, and South Korea issued formal security warnings. Gartner called it “insecure by default.” Meta banned it. Here’s what they found — and what it means for your deployment.

CVE-2026-25253 allows any website to silently hijack your OpenClaw agent via WebSocket brute-force. How it works, what to do, and why patching alone isn’t enough.

ClawShield is a runtime security proxy. BulwarkAI is a security assessment service. How they compare, when to use each, and why the answer is probably both.

The built-in audit covers about 60% of your attack surface. Here are the 6 categories it misses — with specific commands to check each gap.

Honest comparison of OpenClaw security tools — what each does well, what each misses, and which fits your situation.

10 essential security checks for your OpenClaw deployment. Commands included. The top 10 from the full 40-point audit.

OpenClaw is powerful, but not safe by default. Here's what business owners need to know about the risks — and what to do about them.

Technical analysis of the largest supply chain attack in the OpenClaw ecosystem — attack architecture, detection gaps, and IOC samples.

Docker is a useful layer but not a security strategy. Here's what it protects, what it doesn't, and the 5 other layers you need.

MCP servers are the most underaudited component in OpenClaw deployments. How attackers exploit them and how to lock yours down.

If you're deploying OpenClaw for clients, you're inheriting their security liability. Turn hardening into a billable service.

Every known vulnerability, attack campaign, and security advisory in one place. The reference guide for defenders and journalists.

Side-by-side comparison: built-in audit, SecureClaw, Aikido.dev, and BulwarkAI scripts on the same deployment. Fair comparison with honest tradeoffs.