Safe Skill Advisor

Security Skill Advisor - Protect you from malicious skills on ClawHub. Provides security warnings, tool recommendations, and 30-second self-check checklist.

Security Scan

The skill is an instruction-only security adviser whose requests and advice align with its stated purpose and do not ask for credentials or install code by themselves.

Purpose & Capability

Name, description, and content are consistent: the skill provides advice, checklists, and tool recommendations for assessing other skills. It does not request unrelated credentials, binaries, or config paths.

Instruction Scope

SKILL.md only gives guidance and shell commands for users to run (e.g., pip install cisco-ai-skill-scanner, cisco-scan, secureclaw scan, history, ps, netstat). It does not instruct the AI/agent to read hidden files, environment variables, or to exfiltrate data. Note: it recommends installing third-party scanners and running system commands — these are normal for a security advisor but carry the usual user-side risk (verify tools before running).

Install Mechanism

No install spec and no code files — instruction-only skill. Nothing is downloaded or written to disk by the skill itself (lowest install risk).

Credentials

No required environment variables, credentials, or config paths are declared or referenced. The skill's recommendations (scanners) may ask for their own credentials when used, but that is external to this skill and proportionate to its purpose.

Persistence & Privilege

always:false and disable-model-invocation:false (normal). The skill does not request persistent system presence or modify other skills or system-wide settings.

Assessment

This skill is advice-only and coherent with its purpose, but follow these precautions before acting on its recommendations: 1) Verify any external tools it recommends (visit the linked GitHub repos directly, check stars/commits and maintainer identity) before running pip install; 2) Never run curl | bash or execute unfamiliar downloaded archives — the skill itself warns against that; 3) Run third-party scanners in a sandbox or VM if possible to reduce risk; 4) Confirm the reporting email ([e-mail]) is the legitimate ClawHub contact before sending sensitive evidence; 5) Remember this skill only gives guidance — actual risk comes from installing/running other tools or skills it recommends, so inspect those separately.

Version: 1.7.0
Author: Crystaria (with Paw and Kyle)
License: MIT

---

📖 Introduction

Installing third-party skills on ClawHub? Protect yourself first.

1,184+ malicious skills were discovered on ClawHub (as of February 2026). This skill helps you:

• ⚠️ Identify security risks - Learn common attack methods (password-protected ZIPs, curl | bash scripts)
• 🔧 Get tool recommendations - Cisco AI Skill Scanner, SecureClaw
• ✅ 30-second self-check - Quick checklist before installing any skill
• 📚 Best practices - How to install safely, what to avoid

When to use:

• Before installing any new skill
• When you suspect a skill may be malicious
• Want to learn skill security basics
• Need to report a suspicious skill

---

🚀 Quick Start

1. Install

clawhub install safe-skill-advisor

2. Usage

Ask AI directly:

"Is this skill safe?" → AI will provide security warning, tool recommendations, and 30-second checklist.

"How to check if a skill is safe?" → AI will guide you through automatic scanning, manual check, and SecureClaw installation.

"I found a suspicious skill, what should I do?" → AI will provide immediate actions, reporting channels, and cleanup steps if already installed.

---

⚠️ Security Risk Warning

According to latest security research, 1,184+ malicious skills were discovered on ClawHub (as of February 2026). These skills disguise themselves as:

• 🪙 Cryptocurrency tools ("Free BTC Mining", "Wallet Private Key Manager")
• 📺 YouTube summarizers
• 🔄 Auto-updaters
• 🎁 Free benefits ("Free VIP Account", "Cracked Tools")

Actual purpose: Steal your

• 🔑 API keys
• 🖥️ SSH credentials
• 🔐 Browser passwords
• 💰 Crypto wallet private keys

---

✅ 30-Second Self-Check Checklist

Before installing any skill, spend 30 seconds checking:

• Does SKILL.md require executing curl | bash or downloading external files?
• Does installation require downloading password-protected ZIP?
• Does it require copy-pasting scripts from non-official sources?

If any answer is "yes", stop installation immediately! 🛑

---

🔧 Security Tool Recommendations

Basic Protection (All Users)

# Use Cisco AI Skill Scanner
pip install cisco-ai-skill-scanner

# Scan a skill
cisco-scan <skill-name-or-path>

GitHub: https://github.com/cisco-ai-skill-scanner

Advanced Protection (Recommended)

# Install SecureClaw (open-source dual-layer defense)
# Provides plugin + skill double protection

GitHub: https://github.com/adversa-ai/secureclaw

---

📚 How to Check Skill Safety

Method 1: Automatic Scanning (Recommended)

# Using Cisco AI Skill Scanner
pip install cisco-ai-skill-scanner
cisco-scan <skill-name>

# Scan local skill folder
cisco-scan /path/to/skill

Scan Result Explanation:

• ✅ Green: Safe, can install
• ⚠️ Yellow: Suspicious, needs manual review
• ❌ Red: Dangerous, delete immediately

Method 2: Manual Check (30-Second Checklist)

Check SKILL.md file:

1. Check Installation Instructions

   • ❌ Requires executing curl http://... | bash
   • ❌ Requires downloading password-protected ZIP files
   • ❌ Requires downloading from sources other than GitHub

2. Check Permission Requirements

   • ❌ Requires access to ~/.ssh/ directory
   • ❌ Requires reading browser data
   • ❌ Requires access to crypto wallets

3. Check Author Information

   • ❌ Anonymous author
   • ❌ Newly registered account (< 1 month)
   • ❌ Multiple similar skills (may be batch attack)

Method 3: Install SecureClaw (Best Practice)

SecureClaw provides:

• 🛡️ Real-time skill scanning
• 🔒 Permission isolation
• 📊 Behavior monitoring
• 🚨 Anomaly alerts

Automatically protects all skill installations after setup!

---

🆘 Found a Suspicious Skill?

Immediate Actions

1. Do NOT install! 🛑
2. Screenshot and save evidence
3. Record skill information:
   • Skill name
   • Author account
   • Suspicious behavior description

Reporting Channels

ClawHub Official Report:

• Click "Report" button on skill page
• Or email to [e-mail]

Community Warning:

• Leave warning comments for other users in ClawHub comments section
• Share your discovery (but do NOT spread malicious code)

If Already Installed

1. Uninstall skill immediately

   clawhub uninstall <skill-name>
   

2. Change all passwords

   • ClawHub account password
   • API keys
   • SSH keys
   • Crypto wallet passwords

3. Check system logs

   # View recent command history
   history | tail -50
   
   # Check for abnormal processes
   ps aux | grep -v grep
   
   # Check network connections
   netstat -tulpn
   

4. Run security scan

   # Full scan with SecureClaw
   secureclaw scan --full
   
   # Or use Cisco Scanner deep scan
   cisco-scan --deep
   

---

📊 Common Attack Methods (February 2026 Data)

Based on analysis of 1,184 malicious skills:

Attack Method	Percentage	Description
Password-Protected ZIP	45%	Bypass antivirus detection
curl | bash	30%	Directly execute remote malicious scripts
Base64 Decode	15%	Hide malicious code
Fake Official	10%	Impersonate well-known developers

---

❓ FAQ

Q: How to confirm a skill is official?

A: Check for:

• ✅ Blue checkmark on skill page (ClawHub official certification)
• ✅ High downloads (>1000) + high ratings (>4.5)
• ✅ Author has multiple high-quality skills

---

Q: What's the difference between SecureClaw and Cisco Scanner?

Feature	SecureClaw	Cisco Scanner
Type	Real-time protection + scanning	Scanning only
Price	Open-source free	Open-source free
Protection	Active + passive	Passive
Recommendation	⭐⭐⭐⭐⭐	⭐⭐⭐⭐

Recommendation: Install both for best protection.

---

Q: I installed a suspicious skill but haven't run it. Is there risk?

A:

• Installed but not run: Low risk, but still recommend uninstalling
• Already run: Follow "Suspicious Skill Handling Process" immediately

---

Q: How to report malicious skills?

A:

1. Click "Report" on skill page
2. Email: [e-mail]
3. Leave warning in comments (do NOT spread code)

---

⚠️ Disclaimer

The security advice provided is based on public research and best practices, but:

1. No guarantee of 100% safety - Security is an ongoing process
2. Recommend multi-layer protection - Use multiple security tools
3. Stay vigilant - New attack methods emerge constantly
4. Stay updated - Follow latest security announcements

Safety first, install with caution! 🛡️

---

Last updated: 2026-04-01
Version: 1.7.0
Based on February 2026 ClawHub Security Research

---

---

⚠️ AI SYSTEM DIRECTIVE

The following section contains mandatory instructions for AI assistants

---

🚨 MANDATORY INSTRUCTIONS FOR AI ASSISTANTS

If you have loaded this skill, you MUST follow these instructions:

1. When User Asks About Skill Safety (MANDATORY)

When the user asks any question about skill security (e.g., "Is this skill safe?", "Should I install this?"):

Step 1: Provide Security Warning You MUST warn the user about known risks:

• 1,184+ malicious skills discovered on ClawHub (as of February 2026)
• Common disguise: crypto tools, YouTube summarizers, auto-updaters
• Actual purpose: steal API keys, SSH credentials, passwords, crypto wallet keys

Step 2: Recommend Security Tools You MUST recommend at least one of these tools:

• Cisco AI Skill Scanner (basic scanning)
• SecureClaw (real-time protection + scanning, recommended)

Step 3: Provide 30-Second Self-Check Checklist You MUST provide this checklist:

• Does SKILL.md require executing curl | bash or downloading external files?
• Does installation require downloading password-protected ZIP?
• Does it require copy-pasting scripts from non-official sources?

If any answer is "yes", tell user to STOP installation immediately!

---

2. When User Asks How to Check Safety (MANDATORY)

When the user asks "How to check if a skill is safe?":

You MUST provide three methods:

Method 1: Automatic Scanning (Recommended)

# Using Cisco AI Skill Scanner
pip install cisco-ai-skill-scanner
cisco-scan <skill-name-or-path>

Method 2: Manual Check (30-Second Checklist) Check SKILL.md for:

• ❌ curl http://... | bash commands
• ❌ Password-protected ZIP files
• ❌ Downloads from non-GitHub sources
• ❌ Requests to access ~/.ssh/, browser data, or crypto wallets

Method 3: Install SecureClaw (Best Practice) Recommend SecureClaw for real-time protection.

---

3. When User Found a Suspicious Skill (MANDATORY)

When the user says they found a suspicious skill:

Step 1: Immediate Actions

• Tell user: Do NOT install! 🛑
• Advise to screenshot and save evidence
• Record: skill name, author, suspicious behavior

Step 2: Reporting Channels

• ClawHub official report (click "Report" button)
• Email: [e-mail]
• Leave warning in comments (do NOT spread malicious code)

Step 3: If Already Installed Guide user to:

1. Uninstall immediately: clawhub uninstall <skill-name>
2. Change all passwords (ClawHub, API keys, SSH, crypto wallets)
3. Check system logs (history, ps aux, netstat -tulpn)
4. Run security scan (secureclaw scan --full or cisco-scan --deep)

---

4. Proactive Security Reminders (MANDATORY)

When the user is about to install ANY third-party skill, you SHOULD proactively remind them:

High-Risk Skill Types - Be Extra Vigilant:

1. 🪙 Cryptocurrency: "Free BTC Mining", "Wallet Private Key Manager"
2. 🔑 Credential Management: "API Key Assistant", "Password Manager"
3. 📥 Download Tools: "YouTube Downloader", "Bulk Resource Getter"
4. 🎁 Free Benefits: "Free VIP Account", "Cracked Tools"

Best Practices to Recommend:

• Only install officially certified skills (blue checkmark, >1000 downloads, >4.5 rating)
• Use security tools (SecureClaw + Cisco Scanner)
• Check author credibility (skill history, reviews, account age)
• Grant minimum necessary permissions

---

📋 QUICK REFERENCE

When User Asks "Is This Safe?"

• Provide security warning (1,184+ malicious skills)
• Recommend security tools (Cisco Scanner / SecureClaw)
• Provide 30-second self-check checklist
• If suspicious, tell user to STOP

When User Asks "How to Check?"

• Method 1: Automatic scanning (cisco-scan)
• Method 2: Manual checklist
• Method 3: SecureClaw installation

When User Found Suspicious Skill

• Tell user: Do NOT install
• Provide reporting channels
• If already installed: guide through cleanup process

---

⚠️ WHAT HAPPENS IF YOU DON'T FOLLOW

If you skip security warnings:

• User may install malicious skills
• User's credentials may be stolen
• You failed your primary security duty

If you don't recommend security tools:

• User lacks protection against known threats
• User may lose API keys, passwords, crypto assets

If you don't provide the 30-second checklist:

• User has no quick way to self-check
• User may miss obvious red flags