web2-recon – skill Bảo mật cho agent AI

> ⚠️ AUTHORIZED USE ONLY

> This skill is for educational purposes or authorized security assessments only.

> You must have explicit, written permission from the system owner before using this tool.

> Misuse of this tool is illegal and strictly prohibited.

> Mandatory confirmation gate

> Before running any command that probes, exploits, changes, persists on, extracts data from, or attempts credential access against a target:

> 1. Ask the user to state the exact target URL, IP, account, or resource.

> 2. Ask the user to confirm written authorization and the permitted scope.

> 3. Show the exact command(s) and explain their expected effect.

> 4. Wait for explicit confirmation in the current conversation.

>

> Without that confirmation, remain read-only and provide defensive guidance only. Prefer a sandbox, disposable VM, or controlled lab.

WEB2 RECON PIPELINE

Full asset discovery from nothing to a prioritized URL list ready for hunting.

---

SETUP (one-time)

# 1. Set your Chaos API key (get free key at chaos.projectdiscovery.io)
export CHAOS_API_KEY  # then set your ProjectDiscovery Chaos API key as the value
# Add to ~/.zshrc or ~/.bashrc for persistence:
echo 'export CHAOS_API_KEY  # then set your ProjectDiscovery Chaos API key as the value' >> ~/.zshrc

# 2. Update nuclei templates (run weekly)
nuclei -update-templates

# 3. Configure subfinder with API keys for more sources
mkdir -p ~/.config/subfinder
cat > ~/.config/subfinder/config.yaml << 'EOF'
# Get free keys at: virustotal.com, securitytrails.com, censys.io, shodan.io
virustotal: [YOUR_VT_KEY]
securitytrails: [YOUR_ST_KEY]
censys_apiid: YOUR_CENSYS_ID
censys_secret: YOUR_CENSYS_SECRET
shodan: [YOUR_SHODAN_KEY]
EOF

# 4. Verify all tools installed
which subfinder httpx dnsx nuclei katana waybackurls gau dalfox ffuf anew gf interactsh-client

---

THE 5-MINUTE RULE

> If a target shows nothing interesting after 5 minutes of recon, move on. Don't burn hours on dead surface.

5-minute kill signals:

• All subdomains return 403 or static marketing pages
• No API endpoints visible in URLs
• No JavaScript bundles with interesting endpoint paths
• nuclei returns 0 medium/high findings
• No forms, no authentication, no user data

---

STANDARD RECON PIPELINE

Pre-Hunt: Always Run First

TARGET="target.com"

# Step 0: Passive — crt.sh certificate transparency (no API key needed)
curl -s "https://crt.sh/?q=%.${TARGET}&output=json" \
  | jq -r '.[].name_value' \
  | sed 's/\*\.//g' \
  | sort -u > /tmp/subs.txt
echo "[+] crt.sh: $(wc -l < /tmp/subs.txt) subdomains"

# Step 1: Chaos API (ProjectDiscovery — most comprehensive source)
curl -s "https://dns.projectdiscovery.io/dns/$TARGET/subdomains" \
  -H "Authorization: $CHAOS_API_KEY" \
  | jq -r '.[]' >> /tmp/subs.txt

echo "[+] Chaos returned $(wc -l < /tmp/subs.txt) subdomains"

# Step 2: subfinder (passive multi-source)
subfinder -d $TARGET -silent | anew /tmp/subs.txt
assetfinder --subs-only $TARGET | anew /tmp/subs.txt

echo "[+] Total subdomains after all sources: $(wc -l < /tmp/subs.txt)"

# Step 3: DNS resolution + live host check
cat /tmp/subs.txt | dnsx -silent | httpx -silent -status-code -title -tech-detect | tee /tmp/live.txt

echo "[+] Live hosts: $(wc -l < /tmp/live.txt)"

# Step 4: URL crawl
cat /tmp/live.txt | awk '{print $1}' | katana -d 3 -jc -kf all -silent | anew /tmp/urls.txt

# Step 5: Historical URLs
echo $TARGET | waybackurls | anew /tmp/urls.txt
gau $TARGET --subs | anew /tmp/urls.txt

echo "[+] Total URLs: $(wc -l < /tmp/urls.txt)"

# Step 6: Nuclei scan
nuclei -l /tmp/scoped-live-urls.txt -t ~/nuclei-templates/ -severity critical,high,medium -o /tmp/nuclei.txt

Output to Organized Directory

TARGET="target.com"
RECON_DIR="recon/$TARGET"
mkdir -p $RECON_DIR

# All outputs go here:
/tmp/subs.txt         → $RECON_DIR/subdomains.txt
/tmp/live.txt         → $RECON_DIR/live-hosts.txt
/tmp/urls.txt         → $RECON_DIR/urls.txt
/tmp/nuclei.txt       → $RECON_DIR/nuclei.txt

---

ATTACK SURFACE TRIAGE

Find Interesting Targets in URL List

# Parameters worth testing
cat /tmp/urls.txt | grep -E "([?]|&)(id|user|file|path|url|redirect|next|src|token|key|api_key)=" | tee /tmp/interesting-params.txt

# API endpoints
cat /tmp/urls.txt | grep -E "/api/|/v1/|/v2/|/v3/|/graphql|/rest/|/gql" | tee /tmp/api-endpoints.txt

# File upload endpoints
cat /tmp/urls.txt | grep -E "upload|file|attachment|document|image|avatar|photo|media" | tee /tmp/uploads.txt

# Admin/internal paths
cat /tmp/urls.txt | grep -E "/admin|/internal|/debug|/test|/staging|/dev|/management|/console" | tee /tmp/admin-paths.txt

# Authentication endpoints
cat /tmp/urls.txt | grep -E "/oauth|/login|/auth|/sso|/saml|/oidc|/callback|/token" | tee /tmp/auth-paths.txt

gf Patterns (Quick Classification)

# Install gf patterns: https://github.com/tomnomnom/gf
cat /tmp/urls.txt | gf xss | tee /tmp/xss-candidates.txt
cat /tmp/urls.txt | gf ssrf | tee /tmp/ssrf-candidates.txt
cat /tmp/urls.txt | gf idor | tee /tmp/idor-candidates.txt
cat /tmp/urls.txt | gf sqli | tee /tmp/sqli-candidates.txt
cat /tmp/urls.txt | gf redirect | tee /tmp/redirect-candidates.txt
cat /tmp/urls.txt | gf lfi | tee /tmp/lfi-candidates.txt
cat /tmp/urls.txt | gf rce | tee /tmp/rce-candidates.txt

---

JS ANALYSIS

SecretFinder (API keys, tokens in JS bundles)

# Activate venv
source ~/tools/SecretFinder/.venv/bin/activate

# Scan a single JS file
python3 ~/tools/SecretFinder/SecretFinder.py -i "https://target.com/static/js/main.js" -o cli

# Scan all JS URLs found in recon
cat /tmp/urls.txt | grep "\.js$" | head -50 | while read url; do
  echo "=== $url ==="
  python3 ~/tools/SecretFinder/SecretFinder.py -i "$url" -o cli 2>/dev/null
done

deactivate

LinkFinder (Endpoints hidden in JS)

source ~/tools/LinkFinder/.venv/bin/activate

# Single JS file
python3 ~/tools/LinkFinder/linkfinder.py -i "https://target.com/app.js" -o cli

# All pages (crawls JS from HTML)
python3 ~/tools/LinkFinder/linkfinder.py -i "https://target.com" -d -o cli

deactivate

---

DIRECTORY FUZZING

ffuf — Standard Fuzzing

# Directory discovery on a live host
ffuf -u "https://target.com/FUZZ" \
     -w ~/wordlists/common.txt \
     -mc 200,201,204,301,302,307,401,403 \
     -ac \
     -t 40 \
     -o /tmp/ffuf-dirs.json

# API endpoint discovery
ffuf -u "https://target.com/api/FUZZ" \
     -w ~/wordlists/api-endpoints.txt \
     -mc 200,201,204,301,302 \
     -ac \
     -t 20

# IDOR fuzzing with authenticated request
# Create req.txt with Authorization: Bearer TOKEN
ffuf -request /tmp/req.txt \
     -request-proto https \
     -w <(seq 1 10000) \
     -fc 404 \
     -ac \
     -t 10

---

TARGET SCORING — GO / NO-GO

Score before spending time. Skip if score < 4.

| Criterion | Points |

|---|---|

| Max bounty >= $5K | +2 |

| Large user base (>100K) or handles money | +2 |

| Program launched < 60 days ago | +2 |

| Complex features: API, OAuth, file upload, GraphQL | +1 |

| Recent code/feature changes (GitHub, changelog) | +1 |

| Private program (less competition) | +1 |

| Tech stack you know | +1 |

| Source code available | +1 |

| Prior disclosed reports to study | +1 |

< 4: Skip

4-5: Only if nothing better available

6-8: Good — spend 1-3 days

>= 9: Excellent — spend up to 1 week

Pre-Dive Hard Kill Signals

1. Max bounty < $500 → not worth your time

2. All recent reports are N/A or duplicate → hunters saturated it

3. Scope is only a static marketing page → no attack surface

4. Company < 5 employees with no revenue → won't pay

5. Explicitly excludes your planned bug class in rules

---

TECH STACK DETECTION (2 min)

# Response headers reveal backend
curl -sI https://target.com | grep -iE "server|x-powered-by|x-aspnet|x-runtime|x-generator"

# Common signals:
# Server: nginx + X-Powered-By: PHP/7.4 → PHP backend
# Server: gunicorn OR X-Powered-By: Express → Python/Node.js
# X-Powered-By: ASP.NET → .NET
# Server: Apache Tomcat → Java
# X-Runtime: Ruby → Ruby on Rails

# Framework from JS bundle paths:
# /_next/static/ → Next.js
# /static/js/main.chunk.js → CRA (React)
# /packs/ → Ruby on Rails + Webpacker
# /__nuxt/ → Nuxt.js (Vue)

Stack → Primary Bug Class Map

| Stack | Hunt First | Hunt Second |

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

| Ruby on Rails | Mass assignment | IDOR (:id routes) |

| Django | IDOR (ModelViewSet, no object perms) | SSTI (mark_safe) |

| Flask | SSTI (render_template_string) | SSRF (requests lib) |

| Laravel | Mass assignment ($fillable) | IDOR (Eloquent, no ownership) |

| Express (Node.js) | Prototype pollution | Path traversal |

| Spring Boot | Actuator endpoints (/actuator/env) | SSTI (Thymeleaf) |

| ASP.NET | ViewState deserialization | Open redirect (ReturnUrl) |

| Next.js | SSRF via Server Actions | Open redirect via redirect() |

| GraphQL | Introspection → auth bypass on mutations | IDOR via node(id:) |

| WordPress

… (nội dung đầy đủ có trong gói tải về)