Reconnaissance & OSINT Automation Skill for Claude

Reconnaissance & OSINT Automation

Purpose

Enable Claude to conduct comprehensive reconnaissance and open-source intelligence gathering during authorized security assessments. Claude performs passive and active recon using its native analysis capabilities and orchestrates the included scripts for automation at scale.

> Authorization Required: Always confirm written authorization for the target scope before proceeding. Unauthorized reconnaissance is illegal in most jurisdictions.

Activation Triggers

This skill activates when the user asks about:

• Subdomain enumeration or discovery
• DNS reconnaissance, zone transfers, or DNS record analysis
• OSINT gathering on a domain, organization, or person
• Technology fingerprinting or stack identification
• Port scanning, service detection, or banner grabbing
• Google dorking or advanced search query generation
• WHOIS, certificate transparency, or Shodan queries
• Attack surface mapping or perimeter discovery

Prerequisites

pip install requests dnspython python-whois beautifulsoup4 shodan

Optional enhanced capabilities:

• nmap — Active port scanning
• amass — Advanced subdomain enumeration
• theHarvester — Email and domain harvesting
• Shodan API key — Internet-wide device search
• Censys API key — Certificate and host search

Core Capabilities

1. Passive Reconnaissance (No Direct Target Contact)

When the user asks for passive recon or OSINT:

• WHOIS Analysis — Query domain registration records for registrant, registrar, nameservers, and dates. Flag privacy-protected registrations and registrar patterns.
• Certificate Transparency Logs — Search crt.sh for all certificates issued to the domain and subdomains. Extract SANs (Subject Alternative Names) to discover hidden subdomains.
• DNS Records (Passive) — Enumerate A, AAAA, MX, NS, TXT, SOA, SRV, and CNAME records using public resolvers. Analyze SPF, DKIM, and DMARC for email security posture.
• Search Engine Dorking — Generate targeted dork queries to discover exposed files, login portals, and configuration leaks:
• site:target.com filetype:pdf — Exposed documents
• site:target.com inurl:admin — Admin panels
• site:target.com ext:env OR ext:config — Config files
• "@target.com" site:linkedin.com — Employee enumeration
• "target.com" site:pastebin.com — Credential leaks
• Shodan/Censys Queries — Search for internet-exposed services, open ports, banners, and vulnerabilities associated with the target's IP ranges.
• Git/Code Repository Search — Search GitHub/GitLab for leaked credentials, API keys, and internal information:
• org:targetorg api_key
• filename:.env target.com
• "target.com" password

2. Subdomain Enumeration

When the user asks to enumerate subdomains:

• Certificate Transparency — Extract all SANs from crt.sh/Censys certificates (most effective passive method)
• DNS Brute-Force — Run subdomain_enum.py against the common wordlist in resources/
• Wildcard Detection — Query random subdomains to detect wildcard DNS responses and filter false positives
• Resolution Validation — Resolve all candidates to IP addresses; discard NXDOMAINs
• HTTP Probing — Check which subdomains respond on ports 80/443; identify web applications
• Infrastructure Grouping — Group discovered subdomains by IP/ASN to map cloud vs. on-prem assets

Output format for subdomain findings:

Target: example.com
Discovery Method: CT Logs + DNS Brute-Force
Discovered: 47 subdomains

LIVE SUBDOMAINS:
  admin.example.com       → 203.0.113.10  [HTTP 200] [nginx/1.18]
  dev.example.com         → 203.0.113.11  [HTTP 302 → /login]
  api.example.com         → 203.0.113.12  [HTTP 200] [cloudflare]
  internal.example.com    → 10.0.0.5      [No public response — internal?]

INFRASTRUCTURE CLUSTERS:
  203.0.113.10-15 → AS12345 (Company Hosting)
  Cloudflare CDN → 7 subdomains proxied

3. Active Port Scanning & Service Detection

When the user asks to scan ports or detect services:

• Define scan scope (host, subnet, CIDR range) and confirm authorization
• Select scan technique: SYN scan (requires root), connect scan (no root), or stealth options
• Run top-1000 ports first, then targeted service ports
• Perform service version detection (-sV) on all open ports
• Run OS fingerprinting (-O) if authorized
• Grab banners from discovered services
• Flag services with known vulnerabilities based on version data

Provide Nmap commands ready to run:

# Quick discovery
nmap -sn 203.0.113.0/24

# Top 1000 TCP ports with service detection
nmap -sV -sC --top-ports 1000 -oA scan_results 203.0.113.10

# Full port scan with script engine
nmap -sV -sC -p- -T4 -oA full_scan 203.0.113.10

4. DNS Reconnaissance

When the user asks for DNS analysis:

• Enumerate all record types: A, AAAA, MX, NS, TXT, SOA, SRV, CNAME, PTR
• Zone Transfer Attempt (AXFR) — Try against all discovered nameservers:

   dig AXFR @ns1.example.com example.com

• Email Security Analysis:
• SPF: Check for ~all (softfail) or ?all (neutral) — both are weak
• DMARC: Missing DMARC = zero enforcement; p=none = monitoring only
• DKIM: Check selector existence and key strength
• Reverse DNS — PTR lookups on all discovered IPs to find additional hostnames
• DNS History — Check SecurityTrails or PassiveDNS for historical DNS records that may reveal old infrastructure

Flag these misconfigurations:

• Zone transfer allowed → Exposes full DNS zone
• No DMARC record → Email spoofing possible
• SPF +all → Any server can send as this domain
• DNSSEC not configured → DNS cache poisoning risk

5. Technology Fingerprinting

When the user asks to fingerprint technology:

• Analyze HTTP response headers:
• Server: → Web server and version
• X-Powered-By: → Application framework
• Set-Cookie: names → Session framework (PHPSESSID=PHP, JSESSIONID=Java)
• X-Generator: / X-WordPress-Cache: → CMS
• Examine HTML source for meta tags, script includes, CSS frameworks
• Check /robots.txt, /sitemap.xml, /.well-known/ for framework leaks
• Test common CMS paths (/wp-admin/, /administrator/, /wp-json/)
• Analyze JavaScript files for version strings
• Detect WAF presence through header analysis and response behavior
• Identify CDN (Cloudflare, Akamai, Fastly, CloudFront) via IP ranges and headers

Technology stack report format:

URL: https://example.com

WEB SERVER:    nginx/1.18.0 (Ubuntu)
APPLICATION:   WordPress 6.4.2
LANGUAGE:      PHP 8.1
DATABASE:      MySQL (inferred from wp-config patterns)
CDN/WAF:       Cloudflare
JS LIBRARIES:  jQuery 3.6.0, Bootstrap 5.3
TLS:           TLS 1.3, ECDHE-RSA-AES256-GCM-SHA384

NOTABLE HEADERS:
  ✗ Missing: X-Content-Type-Options
  ✗ Missing: X-Frame-Options
  ✗ Missing: Content-Security-Policy
  ✓ Present: Strict-Transport-Security

6. OSINT Correlation & Reporting

When the user asks to correlate OSINT findings:

• Cross-reference IP ranges across WHOIS, ASN lookups, and cloud provider IP lists
• Map employee data (LinkedIn/email patterns) to organizational structure
• Correlate exposed credentials from paste sites against discovered email formats
• Identify third-party services and SaaS platforms in use (via DNS records, JS imports)
• Build infrastructure map showing relationships between assets

Output Standards

Every recon engagement should produce a structured report:

# Reconnaissance Report — [Target]
Date: [Date] | Scope: [Authorized Scope] | Analyst: [Name]

## Executive Summary
[2-3 sentence overview of key findings]

## Discovered Assets
- Subdomains: N found, N live
- IP Ranges: [CIDRs]
- Open Services: [Top findings]
- Technologies: [Stack summary]

## Key Findings
1. [High-impact finding with evidence]
2. [Medium-impact finding]
...

## Attack Surface Summary
[Map of entry points for follow-on testing]

## Recommended Next Steps
- Feed live web apps → Skill 09 (Web Security)
- Feed discovered services → Skill 02 (Vulnerability Scanner)
- Feed cloud assets → Skill 10 (Cloud Security)

Script Reference

subdomain_enum.py

# Passive CT log enumeration
python scripts/subdomain_enum.py --domain target.com --passive-only --output results.json

# Active brute-force with custom wordlist
python scripts/subdomain_enum.py --domain target.com --wordlist resources/common_subdomains.txt --threads 20 --output results.json

dns_recon.py

# Full DNS reconnaissance
python scripts/dns_recon.py --domain target.com --output dns_report.json

# Check zone transfer vulnerability
python scripts/dns_recon.py --domain target.com --check-zone-transfer

tech_fingerprint.py

# Single URL analysis
python scripts/tech_fingerprint.py --url https://target.com --output tech_report.json

# Batch URL fingerprinting
python scripts/tech_fingerprint.py --urls urls.txt --output tech_report.json

Skill Integration

| Next Step | Condition | Target Skill |

|-----------|-----------|--------------|

| Vulnerability assessment | Live services discovered | → Skill 02 |

| Web application testing | Web apps found | → Skill 09 |

| Cloud asset auditing | Cloud-hosted assets found | → Skill 10 |

| Network traffic analysis | PCAP capture available | → Skill 08 |

| IOC correlation | Suspicious infrastructure found | → Skill 06 |

References

• OWASP Testing Guide — Information Gathering
• MITRE ATT&CK — Reconnaissance (TA0043)
• Certificate Transparency — crt.sh
• Shodan Search Engine
• DNSDumpster
• Google Hacking Database (GHDB)

v3.0 Enhancements (2026 Update)

Sharper, more current recon tradecraft:

• Subdomain takeover detection — for every discovered subdomain with a dangling CNAME (pointing to an unclaimed S3/Azure/GitHub Pages/Heroku/Fastly target), flag it as a takeover candidate and identify the fingerprint of the orphaned service.
• Cloud asset discovery — enumerate public S3/GCS/Azure Blob buckets and storage by permutation of the org name; check for listable/world-readable buckets (use cloud_enum-style logic). Map ASN → CIDR → cloud-provider attribution.
• JA4+ / favicon hashing — pivot on favicon mmh3 hashes and JARM/JA4S server fingerprints in Shodan/Censys to find sibling infrastructure behind CDNs.
• GraphQL & API recon — probe /graphql for introspection; harvest schema, types, and mutations to feed Skill 09.
• Breach & credential intel — correlate discovered email formats against HaveIBeenPwned and public combolists (passive only) to prioritize accounts.
• Source-code & CI leaks — extend dorking to GitHub Actions logs, npm/PyPI package metadata, and *.map source maps that leak internal paths.

Precision rule: always separate *confirmed* assets (resolved + responding) from *candidate* assets (CT-log only) in output, and tag the discovery source per asset so findings are reproducible.

Added references

• can-i-take-over-xyz — subdomain takeover fingerprints
• JA4+ network fingerprinting