Agent Skills: HackerOne Bug Bounty Hunting

HackerOne Bug Bounty Hunting

Automates HackerOne workflows: scope parsing → parallel testing → PoC validation → submission reports.

Quick Start

1. Input: HackerOne program URL or CSV file
2. Parse scope and program guidelines
3. Deploy Pentester agents in parallel (one per asset)
4. Validate PoCs (poc.py + poc_output.txt required)
5. Generate HackerOne-formatted reports

Workflows

Option 1: HackerOne URL

- [ ] Fetch program data and guidelines
- [ ] Download scope CSV
- [ ] Parse eligible assets
- [ ] Deploy agents in parallel
- [ ] Validate PoCs
- [ ] Generate submissions

Option 2: CSV File

- [ ] Parse CSV scope file
- [ ] Extract eligible_for_submission=true assets
- [ ] Collect program guidelines
- [ ] Deploy agents
- [ ] Validate and generate reports

Scope CSV Format

Expected columns:

• identifier - Asset URL/domain
• asset_type - URL, WILDCARD, API, CIDR
• eligible_for_submission - Must be "true"
• max_severity - critical, high, medium, low
• instruction - Asset-specific notes

Use tools/csv_parser.py to parse.

Agent Deployment

Pentester Agent per asset:

• Passes program-specific guidelines
• Tests all vulnerability types
• Returns validated findings with PoCs

Parallel Execution:

• 10 assets = 10 Pentester agents
• Each spawns 30+ specialized agents
• Total: 300+ concurrent tests
• Time: 2-4 hours vs 20-40 sequential

PoC Validation (CRITICAL)

Every finding MUST have:

1. poc.py - Executable exploit script
2. poc_output.txt - Timestamped execution proof
3. workflow.md - Manual steps (if applicable)
4. Evidence screenshots/videos

Experimentation: Test edge cases, verify impact, document failures.

Report Format

Required sections (HackerOne standard):

1. Summary (2-3 sentences)
2. Severity (CVSS + business impact)
3. Steps to Reproduce (numbered, clear)
4. Visual Evidence (screenshots/video)
5. Impact (realistic attack scenario)
6. Remediation (actionable fixes)

Use tools/report_validator.py to validate.

Output Structure

Per OUTPUT.md - Bug Bounty format:

outputs/<program>/
├── findings/
│   ├── finding-001/
│   │   ├── report.md           # HackerOne report
│   │   ├── poc.py              # Validated PoC
│   │   ├── poc_output.txt      # Proof
│   │   └── workflow.md         # Manual steps
├── reports/
│   ├── submissions/
│   │   ├── H1_CRITICAL_001.md  # Ready to submit
│   │   └── H1_HIGH_001.md
│   └── SUBMISSION_GUIDE.md
└── evidence/
    ├── screenshots/
    └── http-logs/

Program Selection

High-Value:

• New programs (< 30 days)
• Fast response (< 24 hours)
• High bounties (Critical: $5,000+)
• Large attack surface

Avoid:

• Slow response (> 1 week)
• Low bounties (Critical: < $500)
• Overly restrictive scope

Critical Rules

MUST DO:

• Validate ALL PoCs before reporting
• Sanitize sensitive data
• Test only eligible_for_submission=true assets
• Follow program-specific guidelines
• Generate CVSS scores

NEVER:

• Report without validated PoC
• Test out-of-scope assets
• Include real user data
• Cause service disruption

Quality Checklist

Before submission:

• [ ] Working PoC with poc_output.txt
• [ ] Accurate CVSS score
• [ ] Step-by-step reproduction
• [ ] Visual evidence
• [ ] Impact analysis
• [ ] Remediation guidance
• [ ] Sensitive data sanitized

Tools

• tools/csv_parser.py - Parse HackerOne scope CSVs
• tools/report_validator.py - Validate report completeness
• /pentest skill - Core testing functionality
• Pentester agent - Orchestrates testing

Integration

Uses /pentest skill and Pentester agent. Follows OUTPUT.md for submission format.

Common Rejections

Out of Scope: Check eligible_for_submission=true Cannot Reproduce: Validate PoC, include poc_output.txt Duplicate: Search disclosed reports, submit quickly Insufficient Impact: Show realistic attack scenario

Usage

/hackerone <program_url_or_csv_path>