Skip to content

0xGhostCAT/claude-ai-cyber-security-skills

Repository files navigation

🎯 Claude CyberSecurity Skills

Claude Code Skills + for HackerOne & Bugcrowd hunters. Recon → Hunt → Validate → Report. End-to-end.

License: MIT Claude Code Bug Bounty


What is this?

Claude CyberSecurity Skills transforms Claude Code into a bug bounty hunting partner that knows the workflow, the tools, the payloads, and the platforms.

Most security repos give you a tool list or a methodology PDF. This gives you 30 production-grade Skills that Claude auto-loads when you describe a task — each one wired to real tools (subfinder, nuclei, sqlmap, ffuf, dalfox, etc.), with custom payloads, validation gates, and HackerOne/Bugcrowd-specific report templates.

Tell Claude: "hunt this new H1 program at target.com" Claude: chains program-selection → scope-analysis → subdomain-enum → asset-discovery → fingerprinting → suggested vuln vectors → ready-to-fire commands.

📖 For the full deep-dive: see PROJECT_REPORT.md — comprehensive 22-section report covering architecture, all 30 skills, playbooks, arsenal, ROI analysis, and roadmap.


🧭 The Workflow

                  ┌─────────────────┐
                  │  1. PRE-HUNT    │  pick program • parse scope • mind-map
                  └────────┬────────┘
                           │
                  ┌────────▼────────┐
                  │   2. RECON      │  subdomains • assets • JS • params
                  └────────┬────────┘  (continuous monitoring optional)
                           │
                  ┌────────▼────────┐
                  │   3. HUNT       │  15 vuln classes • chained playbooks
                  └────────┬────────┘
                           │
                  ┌────────▼────────┐
                  │  4. VALIDATE    │  7-Question Gate • 4 pre-submit gates
                  └────────┬────────┘
                           │
                  ┌────────▼────────┐
                  │   5. REPORT     │  H1 / Bugcrowd / Intigriti templates
                  └─────────────────┘

📚 The 30 Skills

Phase 1 — Pre-Hunt (3)

# Skill Trigger phrases
01 program-selection "should I hunt this program", "pick H1 target"
02 scope-analysis "parse this scope", "what's in scope"
03 threat-modeling-mindmap "threat model this app", "mind map target"

Phase 2 — Recon (6)

# Skill Trigger phrases
04 subdomain-enum "enumerate subdomains", "subfinder run"
05 asset-discovery "find live hosts", "probe with httpx"
06 fingerprinting "tech stack", "what's running"
07 js-analysis "analyze JS", "find endpoints in JS"
08 content-discovery "fuzz directories", "find parameters"
09 continuous-monitoring "monitor target", "alert on new subdomain"

Phase 3 — Vuln Hunting (15)

# Skill Trigger phrases
10 idor-hunting "test IDOR", "find IDOR"
11 auth-bypass "bypass auth", "forced browsing"
12 ato-chains "account takeover", "ATO chain"
13 xss "test XSS", "CSP bypass"
14 sqli "SQL injection", "sqlmap"
15 ssrf "test SSRF", "cloud metadata"
16 ssti "template injection", "Jinja2 SSTI"
17 file-upload "upload bypass", "file upload"
18 graphql "test GraphQL", "introspection"
19 jwt-attacks "JWT bypass", "alg=none"
20 oauth-oidc "OAuth flaw", "redirect_uri"
21 business-logic "race condition", "logic flaw"
22 subdomain-takeover "subdomain takeover", "dangling DNS"
23 cache-poisoning "cache poisoning", "web cache"
24 http-smuggling "request smuggling", "CL.TE"

Phase 4 — Frontier (3)

# Skill Trigger phrases
25 llm-ai-security "prompt injection", "test LLM", "AI app"
26 mobile-recon-android "decompile APK", "mobile app"
27 cloud-misconfig "S3 bucket", "exposed cloud"

Phase 5 — Output (3)

# Skill Trigger phrases
28 triage-validation "validate finding", "is this reportable"
29 hackerone-reporting "write H1 report", "HackerOne template"
30 bugcrowd-reporting "Bugcrowd report", "VRT mapping"

🚀 Installation

Linux / WSL (recommended)

git clone https://github.com/0xGhostCAT/claude-cybersecurity-skills.git
cd claude-cybersecurity-skills
chmod +x INSTALL.sh
./INSTALL.sh

Windows (native)

git clone https://github.com/0xGhostCAT/claude-cybersecurity-skills.git
cd claude-cybersecurity-skills
.\INSTALL.ps1

Docker (everything pre-installed)

docker build -t claude-cybersecurity-skills -f docker/claude-cybersecurity-skills.Dockerfile .
docker run -it -v $(pwd)/loot:/loot claude-cybersecurity-skills

Install skills into Claude Code

# Global (all projects)
mkdir -p ~/.claude/skills
cp -r skills/* ~/.claude/skills/

# Project-specific
mkdir -p ./.claude/skills
cp -r skills/* ./.claude/skills/

🎬 Quick Start — First Hunt in 5 Minutes

# 1. Pick a target from H1
claude "I got invited to a private H1 program for *.example.com. Help me decide if it's worth hunting."

# 2. Recon
claude "run full recon on example.com using the subdomain-enum and asset-discovery skills"

# 3. Hunt
claude "the JS file at app.example.com/static/main.js exposes /api/v3/user/{id}/orders. Hunt IDOR."

# 4. Validate before reporting
claude "run the 7-Question Gate on my IDOR finding"

# 5. Report
claude "write a HackerOne report for this IDOR. CVSS Critical scope. Frame impact around PII access."

📦 What's Inside

claude-cybersecurity-skills/
├── README.md                   ← you are here
├── INSTALL.sh / INSTALL.ps1    ← install all tools
├── docker/                     ← all-in-one Docker image
├── skills/                     ← 30 SKILL.md files
├── playbooks/                  ← multi-skill workflows
│   ├── new-target-day1.md
│   ├── continuous-recon-pipeline.md
│   ├── ato-chain-hunt.md
│   ├── ssrf-to-rce-aws.md
│   └── llm-app-fullhunt.md
├── arsenal/                    ← shared payloads & resources
│   ├── wordlists/
│   ├── nuclei-templates/       ← custom (not in nuclei-templates repo)
│   ├── sqlmap-tampers/         ← custom WAF bypass tampers
│   ├── xss-payloads/
│   ├── ssrf-payloads/
│   ├── ssti-payloads/
│   ├── file-upload-polyglots/
│   ├── prompt-injection-payloads/
│   └── semgrep-rules/
├── templates/                  ← report templates per platform
│   ├── hackerone-template.md
│   ├── bugcrowd-template.md
│   ├── intigriti-template.md
│   └── cvss-calculator.md
└── docs/
    ├── 7-question-gate.md
    ├── always-rejected-list.md
    ├── conditional-chain-table.md
    └── tool-cheatsheets/

🧰 Tools Installed by INSTALL.sh

Click to expand full tool list (60+ tools)

Subdomain enumeration: subfinder, amass, assetfinder, chaos, findomain, sublist3r Probing & discovery: httpx, dnsx, naabu, masscan, nmap Crawling: katana, waybackurls, gau, hakrawler, gospider Fuzzing: ffuf, feroxbuster, dirsearch, gobuster Parameter discovery: arjun, paramspider, x8 JS analysis: LinkFinder, SecretFinder, JSluice, mantra, getJS Scanning: nuclei, nikto, wapiti Injection: sqlmap, ghauri, NoSQLMap, dalfox, XSStrike, kxss, gxss, SSRFmap, Gopherus, tplmap, commix Auth: jwt_tool, jwtcat, hydra API/GraphQL: kiterunner, graphw00f, clairvoyance, inql Secrets/SAST: trufflehog, gitleaks, semgrep, noseyparker Cloud: prowler, ScoutSuite, cloudsplaining Mobile: apktool, jadx, MobSF LLM testing: garak, promptfoo Utilities: interactsh-client, anew, qsreplace, unfurl, gron, httpx-toolkit


🎓 Learn the Workflow

  1. The 7-Question Gate — answer all 7 before writing any report.
  2. Always-Rejected List — what NOT to submit. Save your N/A ratio.
  3. Conditional Chain Table — when a "low" becomes a "critical" by chaining.
  4. Tool Cheatsheets — one-pager per tool.

⚖️ Legal & Authorization

This arsenal is for authorized testing only. Use exclusively on:

  • HackerOne / Bugcrowd / Intigriti / Immunefi programs you're enrolled in
  • Your own systems
  • Lab environments (HTB, PortSwigger Academy, etc.)
  • CTF competitions

Unauthorized scanning, exploitation, or data access is illegal in most jurisdictions. Offensive skills include explicit authorization gates that confirm scope before running.

You are responsible for legal compliance. The authors accept no liability for misuse.


📜 License

MIT — see LICENSE.


🙏 Credits & Inspiration


🤝 Contributing

PRs welcome. See CONTRIBUTING.md for how to add a skill, payload, or playbook.


Star ⭐ if useful. Hunt ethically. Get paid.

About

30 Claude Code Skills + 60+ integrated tools for HackerOne/Bugcrowd bug bounty hunters

Topics

Resources

License

Contributing

Stars

Watchers

Forks

Releases

No releases published

Packages

 
 
 

Contributors