Security Policy

September 7, 2026 · View on GitHub

Supported Versions

VersionSupported
3.1.xYes
3.0.xSecurity fixes only — upgrade to v3.1
2.0.xNo — end of life, upgrade to v3.1
1.0.xNo — end of life, upgrade to v3.1

⚠️ Antivirus & VirusTotal False Positives

This repository is a cybersecurity skills collection containing security testing tools, payload templates, and exploit development references for authorized penetration testing and education. Antivirus engines may flag certain files — this is expected behavior and represents a false positive.

Why This Happens

The file skills/03-exploit-development/scripts/payload_generator.py contains reverse shell, web shell, and injection payload templates stored as Python string constants. AV engines use signature and heuristic matching against these exact patterns because they resemble actual malware payloads. However:

  • ❌ No code in this repository executes malicious actions autonomously
  • ❌ No hidden backdoors, C2 communication, or data exfiltration exists
  • ❌ No obfuscated or hidden malicious payloads exist
  • ✅ All payloads are inert string templates requiring explicit user action with authorized targets
  • ✅ Every offensive tool includes legal disclaimers requiring written authorization
  • ✅ Offensive skills enforce authorization gates before Claude provides assistance

Affected Files

FileWhy It's FlaggedActual Purpose
skills/03-exploit-development/scripts/payload_generator.pyContains reverse shell & web shell string templatesGenerates payloads for authorized pentesting
skills/16-ai-llm-security/scripts/prompt_injection_tester.pyContains zero-width characters & obfuscated stringsTests AI model resilience against prompt injection

This Is Standard for Security Repositories

Every major cybersecurity toolkit on GitHub triggers the same AV heuristic detections:

What Users Should Do

If your antivirus quarantines files from this repository:

  1. Whitelist the directory — Add the skills directory to your AV exclusion list
  2. Review the source code — Verify the flagged file contains only string templates
  3. Use in isolated environments — Offensive skills should be used in authorized lab/test environments

Windows Defender

On Windows, real-time protection can silently delete flagged files on write or on every OneDrive sync cycle — for example, payload_generator.py being re-deleted each time it was restored or re-synced, with no on-screen warning (only visible via Get-MpThreatDetection). If files in this repo keep disappearing after git restore/git pull/OneDrive sync, add an exclusion before restoring them, in an elevated PowerShell prompt:

Add-MpPreference -ExclusionPath "C:\path\to\Claude-Code-CyberSecurity-Skill"

Then re-run git restore <file> (or re-sync) to bring the quarantined file back.


Reporting a Vulnerability

If you discover a security vulnerability in this project's code or scripts, do NOT open a public issue.

Report it privately via one of:

  1. GitHub Security AdvisoriesSubmit a private report
  2. GitHub private message — Contact the repository maintainer directly

What to Include

  • Description of the vulnerability
  • Affected file(s) and line numbers if applicable
  • Steps to reproduce
  • Potential impact (what an attacker could do)
  • Suggested fix or mitigation (if you have one)

Response Timeline

StageTarget
AcknowledgmentWithin 48 hours
Initial assessmentWithin 7 days
Fix release (critical)Within 30 days
Fix release (medium/low)Within 90 days

Responsible Use

This skill collection is designed for authorized security testing, research, and education only.

Before using any tool or script in this collection, you must:

  1. Have written authorization from the system owner before testing any system you do not personally own
  2. Comply with all applicable laws (Computer Fraud and Abuse Act, Computer Misuse Act, GDPR, etc.)
  3. Operate only within explicitly defined scope (IP ranges, domains, environments)
  4. Use offensive tools only in isolated, controlled environments when testing your own systems
  5. Report vulnerabilities discovered during authorized testing to the affected parties through responsible disclosure

Skills with offensive capabilities (03-exploit-development, 14-red-team-ops, and 22-purple-team, whose live-fire adversary emulation executes real attack techniques) require authorization verification before Claude provides operational assistance. This is enforced in the SKILL.md authorization gates. Skills 16-ai-llm-security and 18-ot-ics-security carry safety gates for the same reason. For purple team, planning, coverage analysis, and tabletop emulation need no live execution and remain available without an active engagement.

Authorized Use Cases

  • Penetration testing with a signed Statement of Work
  • Bug bounty programs (in-scope targets only)
  • CTF competitions
  • Security research in isolated lab environments
  • Defensive security — hardening, detection engineering, incident response

Prohibited Use Cases

  • Unauthorized access to any computer system
  • Targeting systems outside your authorized scope
  • Distributing discovered vulnerabilities without coordinated disclosure
  • Using scripts to harm, disrupt, or spy on individuals or organizations

Scope of This Security Policy

This policy covers:

  • The skill collection scripts and code
  • The SKILL.md instruction files
  • Documentation and configuration templates

This policy does NOT cover:

  • Third-party tools referenced by the skills (Nmap, Volatility, etc.)
  • Systems or networks tested using these skills
  • User modifications to the scripts or SKILL.md files

Vulnerability Disclosure Credits

Responsible reporters will be credited in the release notes (unless they prefer anonymity).


Back to Main Repository