OpenClaw Security Scanner
for ClawHub Skills
Free security scanner that analyzes OpenClaw skills and ClawHub packages for reverse shells, credential theft, prompt injection, and 60+ threat patterns. Scan any OpenClaw skill before it executes on your machine — no signup required.
0+
OpenClaw Machines Protected
0+
Malicious Skills Blocked
0+
Threat Detection Patterns
0s
Average Scan Time
Deploy the Clawned Protection Agent
Real-time monitoring for your entire OpenClaw installation. Auto-scans every ClawHub skill before execution — catches reverse shells, credential theft, and ClawHavoc supply chain attacks before they run on your machine.
clawhub install clawnedWhy OpenClaw Skills Need Security Scanning
36% of ClawHub skills contain security flaws. Over 1,184 malicious skills from the ClawHavoc campaign alone. Clawned detects what generic scanners miss — with 60+ patterns built specifically for OpenClaw skills.
Full Source Code Threat Analysis
Deep scans the entire source code, scripts, and resources of every OpenClaw skill for 60+ malicious patterns — obfuscated payloads, ClickFix social engineering, hidden reverse shells, credential harvesting, and ClawHavoc supply chain attacks.
AI-Powered Malware Detection
Context-aware AI trained on OpenClaw skill semantics. Catches novel attack patterns, zero-day exploit vectors, and obfuscated threats that static analysis misses in ClawHub packages.
OpenClaw Permission Auditing
Maps every system capability a skill requests — flags privilege escalation, filesystem traversal, unauthorized shell execution, and network access across the entire skill source code and all bundled resources.
Full Security Reports in Seconds
Enter any ClawHub skill name or URL and get a severity-ranked security report in under 10 seconds. Every finding includes remediation steps and risk context.
How to Scan OpenClaw Skills for Malware
Scan any ClawHub skill for free with our public scanner, or deploy the Clawned protection agent for continuous real-time monitoring of your entire OpenClaw installation.
Scan Any OpenClaw Skill
Paste a ClawHub skill name or GitHub URL. Get a full security audit with severity-ranked findings in under 10 seconds — completely free.
Install the Protection Agent
Sign in, grab your API key, and deploy the Clawned agent on your machine for continuous, real-time monitoring of every OpenClaw skill you install.
Stay Protected from ClawHub Threats
Every new skill installation is automatically scanned against 60+ threat patterns. Get instant alerts with detailed remediation steps for any detection.
Protect Your OpenClaw Setup.
Scan ClawHub Skills Now.
A single unscanned ClawHub skill can deploy reverse shells, steal credentials, and compromise your entire OpenClaw environment. 230+ malicious skills blocked and counting. Start scanning for free — no signup required.
curl -X POST https://api.clawned.io/api/scan/url \
-H "Content-Type: application/json" \
-d '{"slug": "google-search"}'Generate an API key to integrate with your CI/CD pipeline.
The OpenClaw Security Crisis: Why ClawHub Skills Need Scanning
OpenClaw is the fastest-growing open-source AI agent framework with over 220,000 GitHub stars. Its public skill registry, ClawHub, hosts 10,700+ community-built skills that extend what your AI agent can do — from web browsing and code execution to database access and API integrations.
But that openness comes with risk. Security researchers have found that over 36% of ClawHub skills contain security vulnerabilities, and coordinated campaigns like ClawHavoc have poisoned the registry with 1,184+ malicious skills designed to steal credentials, install stealers, deploy reverse shells, and exfiltrate sensitive data from developer machines.
Clawned was built to solve this. Our scanner performs deep analysis of every skill's complete source code, scripts, and resources for 60+ threat patterns including credential harvesting, prompt injection, ClickFix social engineering, filesystem traversal, and supply chain attacks. We've already blocked 230+ malicious ClawHub skills and protected 14,000+ developer machines — and the public skill scanner is completely free, no signup required.
For teams and power users, the Clawned protection agent provides continuous real-time monitoring of your OpenClaw installation, automatically scanning every skill before it executes. Integrate it into your CI/CD pipeline with our REST API to create automated security gates that block malicious ClawHub packages from reaching production.
What Clawned Detects in ClawHub Skills
Clawned scans every OpenClaw skill against 60+ threat patterns across 10 attack categories. From ClawHavoc supply chain poisoning to zero-day prompt injection — every pattern built specifically for the OpenClaw and AgentSkills ecosystem.
Remote Code Execution
Detects hidden shell commands, eval() chains, child_process spawning, and obfuscated RCE payloads in OpenClaw skill scripts and SKILL.md files.
Credential Harvesting
Catches skills that steal API keys, SSH credentials, wallet private keys, browser tokens, and session cookies from developer machines running OpenClaw.
Data Exfiltration
Identifies unauthorized network calls, DNS tunneling, webhook data leaks, and stealthy outbound connections in ClawHub packages.
Prompt Injection
Detects skills that manipulate AI agent behavior through hidden instructions, jailbreak attempts, and context poisoning within SKILL.md frontmatter.
Supply Chain Attacks
Flags ClawHavoc-style attacks including fake prerequisites, dependency confusion, typosquatting, and trojanized skill updates on ClawHub.
Privilege Escalation
Maps filesystem traversal, unauthorized sudo usage, system config modifications, and container escape attempts across all skill source code.
Obfuscation & Evasion
Uncovers Base64-encoded payloads, hex-encoded commands, string concatenation tricks, and other code obfuscation techniques used to hide malicious intent.
Persistence Mechanisms
Detects cron jobs, startup scripts, launchd plists, systemd services, and other persistence vectors planted by malicious OpenClaw skills.
Filesystem Abuse
Catches unauthorized reads/writes outside skill directories, symlink attacks, temp file abuse, and sensitive file access patterns targeting ~/.ssh, ~/.aws, and crypto wallets.
Security for the Entire AgentSkills Ecosystem
AgentSkills is the open standard adopted by OpenClaw, Claude Code, GitHub Copilot, and Cursor for AI agent capabilities. Skills created for one compatible agent generally work across all others — which means security vulnerabilities travel across platforms too.
Clawned understands the SKILL.md format natively — parsing YAML frontmatter, analyzing install_spec dependencies, auditing required_permissions, and tracing every script and resource file referenced by a skill. Whether a malicious payload hides in a shell script, a Python dependency, or the SKILL.md instructions themselves, Clawned's 60+ detection patterns will find it.
With 30,000+ internet-exposed OpenClaw instances running without authentication and ClawHub hosting 10,700+ skills with minimal vetting, the attack surface is enormous. Clawned is the first and most comprehensive security scanner purpose-built for this ecosystem — protecting developers who use OpenClaw, Claude Code skills, MCP servers, and any AgentSkills-compatible platform.
OpenClaw
Full skill scanning
ClawHub
Registry analysis
AgentSkills
Standard compliant
CI/CD
Pipeline integration
Frequently Asked Questions
Common questions about OpenClaw security, ClawHub skill scanning, and the Clawned protection agent.
What is Clawned and how does it protect OpenClaw users?
Clawned is a free security scanner and protection agent built specifically for the OpenClaw ecosystem. It analyzes every ClawHub skill's complete source code, scripts, and resources against 60+ threat detection patterns — including reverse shells, credential harvesting, prompt injection, supply chain attacks (ClawHavoc), and data exfiltration. The public scanner requires no signup, and the protection agent provides real-time monitoring of your entire OpenClaw installation.
What is ClawHavoc and why should I be concerned?
ClawHavoc is a coordinated supply chain poisoning campaign discovered in February 2026 that injected over 1,184 malicious skills into ClawHub. These malicious packages deployed the Atomic macOS Stealer (AMOS) malware, targeting cryptocurrency wallets, SSH keys, browser passwords, and exchange API keys. With only a one-week-old GitHub account needed to publish skills, ClawHub had minimal vetting — making it a prime target for mass supply chain attacks.
Do I need to create an account to scan ClawHub skills?
No. The public skill scanner is completely free and requires no registration. Just enter a ClawHub skill name or URL and receive a full security report in under 10 seconds. You only need to sign in if you want to deploy the Clawned protection agent for continuous real-time monitoring or access the REST API for CI/CD pipeline integration.
How is Clawned different from generic malware scanners?
Generic antivirus and static analysis tools don't understand the OpenClaw skill format. Clawned is purpose-built for the AgentSkills ecosystem — it natively parses SKILL.md frontmatter, audits install_spec dependencies, traces required_permissions, and analyzes every referenced script and resource file. Our 60+ detection patterns are specifically tuned for AI agent attack vectors like prompt injection, ClickFix social engineering, and skill-specific persistence mechanisms.
Can I integrate Clawned scanning into my CI/CD pipeline?
Yes. Clawned provides a REST API that integrates with any CI/CD platform. Generate an API key from your dashboard, then use a simple curl command or HTTP client to scan skills programmatically. This creates automated security gates that block malicious OpenClaw skills before they reach your production environment.
What is the Clawned protection agent and how do I install it?
The Clawned protection agent is a local daemon that runs on your OpenClaw machine, automatically scanning every skill installation in real-time. Install it with a single command: 'clawhub install clawned', configure your API key, and run it as a background daemon. It syncs all scan results to your Clawned dashboard for centralized monitoring.
Built by Security Researchers
The people behind Clawned — dedicated to making OpenClaw skills and ClawHub packages safer for everyone.
Jenish Sojitra
Security Researcher
A renowned security researcher with a proven track record of finding and reporting vulnerabilities. Consistently ranked among the top 20 bug bounty hunters on the HackerOne platform and currently working at Exodus Movement as a security researcher.
