$npx -y skills add Orizon-eu/claude-code-pentest --skill vuln-chain-composerComposes multi-step exploit chains by correlating vulnerabilities across domains, calculates real impact of chained findings, generates end-to-end PoC scripts, and produces bug bounty ready reports. Use when user asks to "chain vulnerabilities", "compose exploit chain", "correlat
| 1 | # Vuln Chain Composer |
| 2 | |
| 3 | The strategic brain. Correlates individual vulnerabilities into devastating multi-step exploit chains. |
| 4 | |
| 5 | ## Important |
| 6 | |
| 7 | CRITICAL: This skill produces offensive security analysis. Only use with explicit authorization on the target systems. |
| 8 | |
| 9 | ## Instructions |
| 10 | |
| 11 | ### Step 1: Import Findings |
| 12 | Accept vulnerability data from: |
| 13 | 1. JSON output from webapp-exploit-hunter, api-breaker, cloud-pivot-finder |
| 14 | 2. Manual finding descriptions from the user |
| 15 | 3. Attack tree data from attack-path-architect |
| 16 | 4. Mixed sources - consolidate everything |
| 17 | |
| 18 | ```bash |
| 19 | python scripts/import_findings.py --input {findings_dir_or_files} |
| 20 | ``` |
| 21 | |
| 22 | Normalize all findings to a common format: |
| 23 | - Vulnerability type (SQLi, XSS, SSRF, IDOR, etc.) |
| 24 | - Location (domain, URL, parameter) |
| 25 | - Severity (standalone) |
| 26 | - PoC (if available) |
| 27 | - Prerequisites (authentication level, specific conditions) |
| 28 | |
| 29 | ### Step 2: Cross-Domain Correlation |
| 30 | |
| 31 | ```bash |
| 32 | python scripts/correlate.py --findings {normalized_findings} |
| 33 | ``` |
| 34 | |
| 35 | Analyze relationships between findings: |
| 36 | |
| 37 | **Same-Origin Chains:** |
| 38 | - XSS on subdomain A + sensitive cookies scoped to parent domain = session hijack on all subdomains |
| 39 | - SSRF on subdomain B + internal API access = data exfiltration via internal endpoints |
| 40 | - Open redirect on auth endpoint + OAuth callback = token theft |
| 41 | |
| 42 | **Trust-Based Chains:** |
| 43 | - Subdomain takeover + same cookie scope = full session hijack |
| 44 | - CI/CD access + deployment pipeline = production RCE |
| 45 | - Cloud metadata via SSRF + IAM overprivilege = full cloud compromise |
| 46 | |
| 47 | **Credential-Based Chains:** |
| 48 | - SQL injection + password hashes = credential cracking + account takeover |
| 49 | - .env file exposure + database credentials = direct data access |
| 50 | - IDOR on user profile + email exposure = targeted phishing + account takeover |
| 51 | |
| 52 | **Escalation Chains:** |
| 53 | - Low-privilege IDOR + mass assignment = privilege escalation to admin |
| 54 | - Self-XSS + CSRF = weaponized stored XSS affecting other users |
| 55 | - Rate limit bypass + brute-force + OTP bypass = authentication bypass |
| 56 | |
| 57 | ### Step 3: Chain Construction |
| 58 | |
| 59 | ```bash |
| 60 | python scripts/build_chains.py --correlations {correlation_data} |
| 61 | ``` |
| 62 | |
| 63 | For each identified chain: |
| 64 | |
| 65 | 1. **Define the chain narrative**: Clear story from initial access to final impact |
| 66 | 2. **List each step** with: |
| 67 | - Vulnerability exploited |
| 68 | - Specific URL/parameter |
| 69 | - What is gained at this step |
| 70 | - How it enables the next step |
| 71 | 3. **Map dependencies**: What must succeed for the chain to work |
| 72 | 4. **Identify alternatives**: If one step is fixed, is there a bypass? |
| 73 | 5. **Calculate chain feasibility**: Product of individual step probabilities |
| 74 | |
| 75 | ### Step 4: Impact Recalculation |
| 76 | |
| 77 | ```bash |
| 78 | python scripts/calculate_impact.py --chains {chains_file} |
| 79 | ``` |
| 80 | |
| 81 | Recalculate severity based on chain context: |
| 82 | |
| 83 | **Impact Amplifiers:** |
| 84 | - Self-XSS (Low) + CSRF chain = Stored XSS affecting others (High) |
| 85 | - Info disclosure (Low) + credential reuse = Account takeover (Critical) |
| 86 | - SSRF (Medium) + cloud metadata = Full infrastructure access (Critical) |
| 87 | - IDOR (Medium) + PII access + mass enumeration = Data breach (Critical) |
| 88 | |
| 89 | **CVSS Recalculation:** |
| 90 | For each chain, calculate: |
| 91 | - Attack Complexity: Based on number of steps and prerequisites |
| 92 | - Privileges Required: Based on initial access requirements |
| 93 | - User Interaction: Based on whether victim action is needed |
| 94 | - Scope: Changed if chain crosses trust boundaries |
| 95 | - Confidentiality/Integrity/Availability impact of the FINAL outcome |
| 96 | |
| 97 | ### Step 5: PoC Generation |
| 98 | |
| 99 | ```bash |
| 100 | python scripts/generate_chain_poc.py --chain {chain_file} |
| 101 | ``` |
| 102 | |
| 103 | For each confirmed chain, generate: |
| 104 | |
| 105 | 1. **Step-by-step reproduction guide** with screenshots descriptions |
| 106 | 2. **Automated PoC script** (Python) that: |
| 107 | - Executes each step in sequence |
| 108 | - Passes data between steps (tokens, cookies, IDs) |
| 109 | - Validates each step succeeded before continuing |
| 110 | - Generates evidence at each step |
| 111 | 3. **curl command sequence** for manual reproduction |
| 112 | 4. **Video script**: Narrated steps for recording a PoC video |
| 113 | |
| 114 | ### Step 6: Bug Bounty Report Generation |
| 115 | |
| 116 | ```bash |
| 117 | python scripts/generate_report.py --chains {chains_file} --format {platform} |
| 118 | ``` |
| 119 | |
| 120 | Platform-optimized reports for: |
| 121 | - **HackerOne format**: Title, severity, description, steps to reproduce, impact, remediation |
| 122 | - **Bugcrowd format**: Similar with VRT classification |
| 123 | - **Generic format**: Professional pentest report style |
| 124 | |
| 125 | Report structure per chain: |
| 126 | ``` |
| 127 | ## Title |
| 128 | [Compelling, impact-focused title] |
| 129 | |
| 130 | ## Severity |
| 131 | [Recalculated sev |