Security Skills for CLI Agents

A collection of security testing skills built from analyzing 4,000+ paid HackerOne bug bounty reports. These skills can be used with Claude Code, Gemini CLI, or any agent that supports MCP/Skills.

What is this?

This repo contains specialized prompts (skills) that teach AI coding agents how to find real security vulnerabilities. Instead of dumping thousands of bug reports into context, each skill distills the patterns and techniques from hundreds of real-world findings into actionable guidance.

Available Skills

Skill	Description
mitm-find-idor	Find Insecure Direct Object Reference vulnerabilities
mitm-find-auth	Detect authentication and authorization issues
mitm-find-bizlogic	Identify business logic flaws
mitm-find-ssrf	Find Server-Side Request Forgery vulnerabilities
mitm-find-sqli	Detect SQL injection patterns
mitm-find-otp	Find OTP/2FA bypass vulnerabilities
mitm-find-pii	Identify PII exposure issues
mitm-find-secrets	Detect leaked secrets and API keys
mitm-find-callback	Find callback/webhook security issues
mitm-find-checksum	Identify checksum/integrity bypass opportunities
mitm-find-enumerable	Find enumerable endpoints and IDs
mitm-find-insecure	Detect insecure configurations
mitm-find-referer	Find referer-based vulnerabilities
mitm-list-apis	List and analyze captured API endpoints
mitm-subdomains	Analyze subdomain patterns
mitm-security-audit	Run comprehensive security audit
mitm-report	Generate security report

Setup

For Claude Code

Copy skills to your project's .claude/skills/ directory:

mkdir -p .claude/skills
cp *.md .claude/skills/

For Gemini CLI

Copy as commands to .gemini/commands/:

mkdir -p .gemini/commands
cp *.md .gemini/commands/

Usage

1. Start mitmproxy to capture traffic:

   mitmdump -w traffic.mitm --set flow_detail=3 2>&1 | tee log.txt &

2. Configure your browser/app to proxy through localhost:8080

3. Browse the target application to capture traffic

4. Ask your AI agent to analyze:

   Find security issues in example.com
   Check for idor and auth issues
   Run a full security audit
   

How It Works

Rather than overwhelming the AI with raw bug reports, each skill contains:

• High-value patterns extracted from real bounty-winning reports
• Specific grep/regex patterns to search traffic logs
• Testing methodology with curl examples
• Severity ratings and impact assessment
• False positive guidance

Disclaimer

Only use these tools on systems you have explicit permission to test. Unauthorized security testing is illegal. These skills are intended for:

• Authorized penetration testing
• Bug bounty programs where you have permission
• Security research on your own systems
• Educational purposes

Credits

Built by analyzing 4,000+ paid bug bounty reports from HackerOne's public disclosures.