SuperAPI AI Directory
Partner中文AI newsFavorites
Skillssubdomain-takeover

subdomain-takeover

Security & auditing

View repository

Updated 10-16 07:39

About this skill

What this skill does

Detects and verifies subdomain takeover vulnerabilities by identifying dangling DNS CNAME records that point to unclaimed or deprovisioned third-party services. For each CNAME, resolves the target, checks the HTTP response against known vulnerable-service fingerprints, and confirms exploitability by verifying the resource can be claimed. Covers 18+ vulnerable services including GitHub Pages, AWS S3, Heroku, Netlify, Azure (Traffic Manager, Blob, CloudApp), Shopify, Fastly, Ghost, Pantheon, Tumblr, WordPress.com, Cargo Collective, Surge.sh, Bitbucket, Zendesk, Readme.io, and Statuspage.

When to use

  • During recon on a bug bounty program to find high-impact, low-effort findings
  • After subdomain enumeration (e.g. from the subdomain-enum skill) to check each live CNAME
  • When auditing an organization's DNS records for hygiene and dangling entries
  • After a product decommissioning or cloud migration to verify DNS cleanup was complete

Prerequisites

  • dig installed (part of dnsutils on Debian/Ubuntu, bind-utils on RHEL/CentOS): 
    apt install dnsutils   # Debian/Ubuntu
  • curl installed
  • httpx installed (for bulk HTTP probing): 
    go install -v github.com/projectdiscovery/httpx/cmd/httpx@latest
  • One of the following automated scanners (choose one):
    • subjack: 
      go install github.com/haccer/subjack@latest
    • nuclei with takeover templates: 
      go install -v github.com/projectdiscovery/nuclei/v3/cmd/nuclei@latest
nuclei -update-templates
  • Environment variable SECSKILL_TARGET_DOMAIN: target root domain
  • Environment variable SECSKILL_SUBDOMAIN_LIST: path to a file of subdomains (one per line); can be output from the subdomain-enum skill

Inputs

Variable Required Description
SECSKILL_SUBDOMAIN_LIST required Path to file containing subdomains to check (one per line)
SECSKILL_TARGET_DOMAIN required Root domain (used for labeling output files)
SECSKILL_OUTPUT_DIR optional Directory to save results (default: ./output)
SECSKILL_USE_SUBJACK optional Set to true to run subjack (default: false)
SECSKILL_USE_NUCLEI optional Set to true to run nuclei takeover templates (default: false)

Workflow

Step 1: Prepare environment

export TARGET="${SECSKILL_TARGET_DOMAIN:?Set SECSKILL_TARGET_DOMAIN}"
export SUBLIST="${SECSKILL_SUBDOMAIN_LIST:?Set SECSKILL_SUBDOMAIN_LIST}"
export OUTDIR="${SECSKILL_OUTPUT_DIR:-./output}"
mkdir -p "$OUTDIR"

echo "[*] Target domain : $TARGET"
echo "[*] Subdomain list: $SUBLIST ($(wc -l < "$SUBLIST") entries)"
echo "[*] Output dir    : $OUTDIR"

Step 2: Resolve CNAME records for all subdomains

For each subdomain, resolve the CNAME chain to find the ultimate target. A dangling record exists when the CNAME target belongs to a third-party service that no longer has the resource provisioned.

echo "[*] Resolving CNAME records..."
> "$OUTDIR/cnames_${TARGET}.txt"

while IFS= read -r SUB; do
  CNAME=$(dig +short CNAME "$SUB" 2>/dev/null | sed 's/\.$//')
  if [ -n "$CNAME" ]; then
    echo "$SUB -> $CNAME" | tee -a "$OUTDIR/cnames_${TARGET}.txt"
  fi
done < "$SUBLIST"

TOTAL_CNAMES=$(wc -l < "$OUTDIR/cnames_${TARGET}.txt")
echo "[+] CNAMEs found: $TOTAL_CNAMES"
echo "[+] Saved to: $OUTDIR/cnames_${TARGET}.txt"

Step 3: Check HTTP responses for known takeover fingerprints

Probe each CNAME-bearing subdomain over HTTP/HTTPS and look for service-specific error responses that indicate the resource is unclaimed.

echo "[*] Probing CNAME subdomains for takeover fingerprints..."

# Fingerprint database: service name -> response string indicating unclaimed resource
# See REFERENCE.md for the full fingerprint table.
declare -A FINGERPRINTS
FINGERPRINTS["github-pages"]="There isn't a GitHub Pages site here"
FINGERPRINTS["s3"]="NoSuchBucket"
FINGERPRINTS["heroku"]="No such app"
FINGERPRINTS["netlify"]="Not Found - Request ID"
FINGERPRINTS["azure-trafficmanager"]="404 Not Found"
FINGERPRINTS["azure-blob"]="BlobNotFound"
FINGERPRINTS["azure-cloudapp"]="404 Web Site not found"
FINGERPRINTS["shopify"]="Sorry, this shop is currently unavailable"
FINGERPRINTS["fastly"]="Fastly error: unknown domain"
FINGERPRINTS["ghost"]="Failed to resolve DNS for"
FINGERPRINTS["pantheon"]="The gods are wise, but do not know of the site which you seek"
FINGERPRINTS["tumblr"]="There's nothing here"
FINGERPRINTS["wordpress-com"]="Do you want to register"
FINGERPRINTS["cargo-collective"]="404 Not Found"
FINGERPRINTS["surge-sh"]="project not found"
FINGERPRINTS["bitbucket"]="Repository not found"
FINGERPRINTS["zendesk"]="Help Center Closed"
FINGERPRINTS["readme-io"]="Project doesnt exist"
FINGERPRINTS["statuspage"]="Better Uptime"

> "$OUTDIR/vulnerable_${TARGET}.txt"

while IFS= read -r SUB; do
  CNAME=$(dig +short CNAME "$SUB" 2>/dev/null | sed 's/\.$//')
  [ -z "$CNAME" ] && continue

  RESPONSE=$(curl -sk --max-time 10 "https://$SUB" 2>/dev/null || curl -sk --max-time 10 "http://$SUB" 2>/dev/null)
  [ -z "$RESPONSE" ] && continue

  for SERVICE in "${!FINGERPRINTS[@]}"; do
    FP="${FINGERPRINTS[$SERVICE]}"
    if echo "$RESPONSE" | grep -qi "$FP"; then
      echo "[VULNERABLE] $SUB -> $CNAME | Service: $SERVICE | Fingerprint: $FP" \
        | tee -a "$OUTDIR/vulnerable_${TARGET}.txt"
      break
    fi
  done
done < "$SUBLIST"

VULN_COUNT=$(wc -l < "$OUTDIR/vulnerable_${TARGET}.txt")
echo "[+] Potentially vulnerable subdomains: $VULN_COUNT"
echo "[+] Results saved to: $OUTDIR/vulnerable_${TARGET}.txt"

Step 4: Automated scan with subjack (optional)

subjack automates fingerprint detection across all subdomains and checks against an up-to-date fingerprints database.

if [ "${SECSKILL_USE_SUBJACK:-false}" = "true" ]; then
  echo "[*] Running subjack..."
  subjack \
    -w "$SUBLIST" \
    -t 100 \
    -timeout 30 \
    -ssl \
    -c "$(go env GOPATH)/pkg/mod/github.com/haccer/subjack@*/fingerprints.json" \
    -o "$OUTDIR/subjack_${TARGET}.txt" \
    -v
  echo "[+] subjack results saved to: $OUTDIR/subjack_${TARGET}.txt"
else
  echo "[*] Skipping subjack (enable with SECSKILL_USE_SUBJACK=true)"
fi

Step 5: Automated scan with nuclei takeover templates (optional)

nuclei's takeover template pack covers 50+ services and is updated regularly by the ProjectDiscovery community.

if [ "${SECSKILL_USE_NUCLEI:-false}" = "true" ]; then
  echo "[*] Running nuclei takeover templates..."
  nuclei \
    -l "$SUBLIST" \
    -t http/takeovers/ \
    -o "$OUTDIR/nuclei_takeover_${TARGET}.txt" \
    -silent
  echo "[+] nuclei results saved to: $OUTDIR/nuclei_takeover_${TARGET}.txt"
else
  echo "[*] Skipping nuclei (enable with SECSKILL_USE_NUCLEI=true)"
fi

Step 6: Manual verification of candidate subdomains

For each candidate identified in Steps 3–5, manually verify exploitability before reporting. False positives are common — confirm that the specific resource (bucket, GitHub Pages repo, Heroku app name, etc.) is truly unclaimed.

# Manual verification procedure for each candidate:
CANDIDATE_SUB="<subdomain from vulnerable list>"
CANDIDATE_CNAME=$(dig +short CNAME "$CANDIDATE_SUB" | sed 's/\.$//')
echo "[*] Verifying: $CANDIDATE_SUB -> $CANDIDATE_CNAME"

# 1. Confirm CNAME resolution
dig +short CNAME "$CANDIDATE_SUB"

# 2. Check HTTP response and status code
curl -svk "https://$CANDIDATE_SUB" 2>&1 | grep -E "< HTTP|There isn|NoSuch|No such|not found" | head -20

# 3. Identify the service from the CNAME target
echo "$CANDIDATE_CNAME" | grep -oE "(github\.io|s3\.amazonaws\.com|herokuapp\.com|netlify\.app|azurewebsites\.net|trafficmanager\.net|blob\.core\.windows\.net|shopify\.com|fastly\.net|ghost\.io|pantheonsite\.io|tumblr\.com|wordpress\.com|cargocollective\.com|surge\.sh|bitbucket\.io|zendesk\.com|readme\.io|statuspage\.io)"

Service-specific claim verification:

Reference: See REFERENCE.md for the service-specific claim verification table.

Step 7: Attempt claim (authorized testing only)

Only attempt to claim a resource if you have explicit written authorization from the domain owner. Claiming a subdomain without authorization constitutes unauthorized access in most jurisdictions.

# Example: GitHub Pages claim (authorized only)
# 1. Create a GitHub repository matching the expected Pages URL
# 2. Add a simple index.html with a non-harmful proof-of-concept page
# 3. Enable GitHub Pages on the repository
# 4. Verify the subdomain resolves to your proof page

echo "AUTHORIZED TESTING ONLY — do not claim without written permission"
echo "[*] If claiming for PoC: create a benign proof page (e.g. 'Subdomain takeover PoC - <your name>') "
echo "[*] Do NOT host any malicious content on the claimed resource"
echo "[*] Report immediately and release the resource after confirmation"

Step 8: Document affected subdomains and produce report

echo "[*] Generating summary report..."

cat > "$OUTDIR/takeover_report_${TARGET}.md" << EOF
## Subdomain Takeover Report: $TARGET

**Scan date**: $(date +%Y-%m-%d)
**Subdomains scanned**: $(wc -l < "$SUBLIST")
**CNAMEs found**: $TOTAL_CNAMES
**Vulnerable candidates**: $VULN_COUNT

### Confirmed Vulnerable Subdomains

| Subdomain | CNAME Target | Service | Severity | Status |
|-----------|-------------|---------|----------|--------|
$(cat "$OUTDIR/vulnerable_${TARGET}.txt" 2>/dev/null | sed 's/\[VULNERABLE\] //g' | awk -F' -> | Service: | Fingerprint: ' '{print "| "$1" | "$2" | "$3" | High | Unverified |"}')

### Remediation

For each affected subdomain:
1. Remove the dangling DNS CNAME record immediately, OR
2. Re-provision the target service resource if it is still needed
3. Implement DNS record review as part of decommissioning checklists

### CVSS

Subdomain takeover is typically rated **High (CVSS 8.1)**:
- Attack Vector: Network, Attack Complexity: Low, Privileges Required: None
- User Interaction: Required (victim must visit the subdomain)
- Scope: Changed, Confidentiality: High, Integrity: High, Availability: None

Impact increases to **Critical** if:
- The subdomain handles authentication cookies (cookie scope inheritance)
- The subdomain is trusted by CORS policies of the main application
- The subdomain can serve malicious JS loaded by the main application
EOF

echo "[+] Report saved to: $OUTDIR/takeover_report_${TARGET}.md"

Done when

  • All subdomains from the input list have been checked for CNAME records
  • Each CNAME-bearing subdomain has been probed and compared against the fingerprint database
  • Automated scanners (subjack or nuclei) have been run if enabled
  • Each candidate has been manually verified for true exploitability (not just a matching fingerprint)
  • The final report lists confirmed vulnerable subdomains with CNAME target, service, severity, and remediation advice

Failure modes

Symptom Cause Resolution
subjack: command not found Not installed or not in PATH Run go install github.com/haccer/subjack@latest; ensure $(go env GOPATH)/bin is in $PATH
CNAME resolves but HTTP probe returns no response Service is active but on a non-standard port Probe ports 80, 443, 8080; some services use HTTP-only on port 80
Fingerprint matched but resource is actually claimed Stale fingerprint or partial page load Manually browse the URL and attempt to reproduce the claim; check for custom 404 pages mimicking takeover messages
High false positive rate Generic 404 pages matching fingerprints Tune fingerprint strings to be more specific; cross-check with service-specific claim verification in Step 6
nuclei templates not found Templates not downloaded Run nuclei -update-templates first
dig returns NXDOMAIN for subdomain DNS record already removed Mark as remediated; no longer vulnerable

Notes

  • Subdomain takeover is typically rated High severity (CVSS ~8.1) standalone, but can escalate to Critical if the subdomain is in the cookie scope of the root domain, trusted by the main app's CORS policy, or included as a script source — enabling session hijacking or XSS against all users of the main application.
  • GitHub Pages takeovers are the most common and easiest to verify — a 404 with "There isn't a GitHub Pages site here" is highly reliable.
  • AWS S3 bucket takeovers require knowing the region; try us-east-1 first, then enumerate other regions if the initial bucket creation fails.
  • Always release claimed PoC resources immediately after the report is accepted or a reasonable reporting period has elapsed (e.g. 7 days with no response).
  • Feed output from the subdomain-enum skill directly as SECSKILL_SUBDOMAIN_LIST for a complete recon-to-takeover pipeline.
  • Nuclei takeover template reference: https://github.com/projectdiscovery/nuclei-templates/tree/main/http/takeovers
  • subjack reference: https://github.com/haccer/subjack