@aegis-scan/skills 0.4.0 → 0.5.1

package/skills/offensive/airecon-fork/reconnaissance-javascript-analysis/SKILL.md

@@ -0,0 +1,167 @@
+<!-- aegis-local: forked 2026-05-04 from pikpikcu/airecon@9a21453459d87eefb012ea355c79b593d0d3c0cc (MIT-licensed); attribution preserved, see ATTRIBUTION.md -->
+
+# JavaScript Endpoint & Secret Extraction
+
+Use this playbook when analyzing any web application that serves JavaScript files.
+Modern SPAs (React, Vue, Angular, Next.js) expose almost all API routes and sometimes secrets inside JS bundles.
+
+---
+
+## STEP 1 — Collect All JavaScript File URLs
+
+```bash
+# From browser_action output (saved to output/js_files.txt):
+cat output/js_files.txt
+
+# OR: extract from raw HTML if browser_action wasn't used:
+curl -sk https://TARGET/ | grep -oP '(?<=src=")[^"]+\.js[^"]*' | sed 's|^/|https://TARGET/|' | tee output/js_files.txt
+
+# Also check for chunk manifest / lazy-loaded bundles:
+curl -sk https://TARGET/ | grep -oP '(?<=src=")[^"]+' | grep -E '\.(js|chunk)' | tee -a output/js_files.txt
+curl -sk https://TARGET/asset-manifest.json 2>/dev/null | python3 -c "import sys,json; d=json.load(sys.stdin); [print(v) for v in d.values() if '.js' in str(v)]"
+curl -sk https://TARGET/webpack-manifest.json 2>/dev/null | python3 -c "import sys,json; [print(v) for k,v in json.load(sys.stdin).items() if '.js' in str(v)]"
+```
+
+---
+
+## STEP 2 — Download All JS Files
+
+```bash
+mkdir -p output/js_files
+while IFS= read -r url; do
+  [ -z "$url" ] && continue
+  # Resolve relative URLs
+  [[ "$url" != http* ]] && url="https://TARGET${url}"
+  fname=$(echo "$url" | md5sum | cut -d' ' -f1).js
+  curl -sk "$url" -o "output/js_files/$fname" && echo "Downloaded: $url -> $fname"
+done < output/js_files.txt
+echo "Total JS files: $(ls output/js_files/*.js 2>/dev/null | wc -l)"
+```
+
+---
+
+## STEP 3 — Extract API Endpoints
+
+```bash
+# Pattern 1: quoted string paths starting with / or /api
+grep -roh '"\(/[a-zA-Z0-9_/.-]\{2,80\}\)"' output/js_files/ 2>/dev/null \
+  | tr -d '"' | sort -u | grep -E '^/(api|v[0-9]|auth|user|admin|account|order|product|data|graphql)' \
+  | tee output/js_extracted_endpoints.txt
+
+# Pattern 2: single-quoted paths
+grep -roh "'\(/[a-zA-Z0-9_/.-]\{2,80\}\)'" output/js_files/ 2>/dev/null \
+  | tr -d "'" | sort -u | grep -E '^/(api|v[0-9]|auth|user|admin)' \
+  >> output/js_extracted_endpoints.txt
+
+# Pattern 3: template literals with path
+grep -roh '`/[a-zA-Z0-9_/.-]\{2,60\}`' output/js_files/ 2>/dev/null \
+  | tr -d '`' | sort -u >> output/js_extracted_endpoints.txt
+
+# Pattern 4: fetch/axios/http calls (finds dynamic routes)
+grep -roh 'fetch\s*([^)]\{5,120\})' output/js_files/ 2>/dev/null | head -30 >> output/js_extracted_endpoints.txt
+grep -roh 'axios\.[a-z]\+\s*([^)]\{5,120\})' output/js_files/ 2>/dev/null | head -30 >> output/js_extracted_endpoints.txt
+
+# Deduplicate and show results
+sort -u output/js_extracted_endpoints.txt | head -50
+echo "Total unique endpoints found: $(sort -u output/js_extracted_endpoints.txt | wc -l)"
+```
+
+---
+
+## STEP 4 — Scan for Secrets & Hardcoded Credentials
+
+```bash
+# API keys and tokens
+grep -roh 'api[_-]\?key[^"'"'"'`]\{0,10\}["\`'"'"'][A-Za-z0-9_\-]\{16,64\}' \
+  output/js_files/ -i 2>/dev/null | head -20 | tee output/js_secrets.txt
+
+# Auth tokens and secrets
+grep -roh '\b\(secret\|token\|password\|passwd\|auth\|credential\)[^"'"'"'`]\{0,15\}["\`'"'"'][A-Za-z0-9_=+/\-]\{8,100\}' \
+  output/js_files/ -i 2>/dev/null | head -20 >> output/js_secrets.txt
+
+# AWS keys
+grep -roh 'AKIA[A-Z0-9]\{16\}' output/js_files/ 2>/dev/null | head -5 >> output/js_secrets.txt
+grep -roh '"aws[^"]\{0,20\}": *"[A-Za-z0-9/+]\{40\}"' output/js_files/ -i 2>/dev/null >> output/js_secrets.txt
+
+# Internal URLs / backend hosts
+grep -roh 'https\?://[a-zA-Z0-9._-]\{4,80\}' output/js_files/ 2>/dev/null \
+  | grep -v -E 'cdn\.|static\.|fonts\.|analytics\.|google\.|facebook\.' \
+  | sort -u | tee output/js_internal_urls.txt
+
+# Show found secrets
+echo "=== Secrets found ==="
+cat output/js_secrets.txt
+echo "=== Internal URLs ==="
+head -20 output/js_internal_urls.txt
+```
+
+---
+
+## STEP 5 — Next.js / React Router Route Extraction
+
+```bash
+# Next.js: page routes from main-*.js
+grep -roh '"pathname":"[^"]\{1,80\}"' output/js_files/ 2>/dev/null \
+  | grep -oP '(?<=pathname":")[^"]+' | sort -u | tee output/js_nextjs_routes.txt
+
+# React Router: look for route definitions
+grep -roh 'path:\s*["\'"'"'][^"'"'"']\{1,80\}["\'"'"']' output/js_files/ 2>/dev/null \
+  | grep -oP '(?<=path: ["\'"'"'])[^"'"'"']+' | sort -u >> output/js_nextjs_routes.txt
+
+# Angular: routerLink / loadChildren
+grep -roh 'routerLink:\s*["\'"'"'][^"'"'"']\{1,80\}["\'"'"']' output/js_files/ 2>/dev/null | head -20
+grep -roh 'loadChildren.*\.module' output/js_files/ 2>/dev/null | head -10
+
+# Webpack chunk IDs → download extra chunks
+grep -roh '"[0-9]\{1,4\}":"[a-f0-9]\{8,16\}"' output/js_files/ 2>/dev/null \
+  | python3 -c "
+import sys, json, re
+chunks = {}
+for line in sys.stdin:
+    m = re.findall(r'\"(\d+)\":\"([a-f0-9]{8,16})\"', line)
+    chunks.update(m)
+for cid, chash in list(chunks.items())[:20]:
+    print(f'Chunk {cid}: TARGET/static/js/{cid}.{chash}.chunk.js')
+" | tee output/js_webpack_chunks.txt
+```
+
+---
+
+## STEP 6 — Test Discovered Endpoints
+
+```bash
+# Load discovered endpoints and probe each one:
+while IFS= read -r endpoint; do
+  [ -z "$endpoint" ] && continue
+  response=$(curl -sk -o /dev/null -w "%{http_code}" "https://TARGET${endpoint}")
+  [ "$response" != "404" ] && echo "[$response] $endpoint"
+done < output/js_extracted_endpoints.txt | tee output/js_live_endpoints.txt
+
+# Test with authentication cookie (if you have one):
+while IFS= read -r endpoint; do
+  [ -z "$endpoint" ] && continue
+  response=$(curl -sk -b output/cookies.txt -o /dev/null -w "%{http_code}" "https://TARGET${endpoint}")
+  [ "$response" != "404" ] && echo "[$response] $endpoint"
+done < output/js_extracted_endpoints.txt | tee output/js_authed_endpoints.txt
+```
+
+---
+
+## Key Patterns to Look For
+
+**Unauthenticated API routes** — endpoints that return 200 without a cookie/token
+**Admin/internal routes** — `/api/admin`, `/internal/`, `/_/`, `/debug/`
+**IDOR candidates** — routes containing `{id}`, `:id`, `[id]`, or numeric path segments
+**File operations** — `/upload`, `/download`, `/export`, `/import`
+**State-changing ops** — POST/PUT/DELETE endpoints (note them for CSRF/IDOR testing)
+**Hardcoded credentials** — any `password:`, `secret:`, `apiKey:` values in plain text
+
+## Common Frameworks Quick Reference
+
+| Framework | Bundle Pattern | Route Location |
+|-----------|---------------|----------------|
+| Next.js | `_next/static/chunks/` | `pathname:"..."` in main-*.js |
+| React CRA | `static/js/main.*.js` | React Router: `path="..."` |
+| Vue CLI | `js/app.*.js` | vue-router: `path: '...'` |
+| Angular | `main.*.js` | `loadChildren`, `routerLink` |
+| Webpack | `*.chunk.js` | Chunk manifest |

package/skills/offensive/airecon-fork/reconnaissance-js-internal-hostname-intelligence/SKILL.md

@@ -0,0 +1,391 @@
+<!-- aegis-local: forked 2026-05-04 from pikpikcu/airecon@9a21453459d87eefb012ea355c79b593d0d3c0cc (MIT-licensed); attribution preserved, see ATTRIBUTION.md -->
+
+---
+name: js-internal-hostname-intelligence
+description: Deep JavaScript bundle analysis focused on extracting internal hostnames, fleet/cluster naming conventions, obfuscated infrastructure references, and non-API internal service URLs that reveal backend topology
+---
+
+# JavaScript Bundle — Internal Hostname & Infrastructure Intelligence
+
+Standard JS analysis extracts API endpoints and secrets. This skill goes deeper: extracting **internal hostnames** that reveal backend infrastructure topology, fleet naming conventions, internal service mesh references, and obfuscated hostnames (hex-encoded, base64-encoded, or split across variables).
+
+**Why this matters:** Internal hostnames in JS bundles are almost always unintentional leaks. They reveal:
+- Internal service names that are publicly routable (misconfigured cluster ingress)
+- Fleet/Kubernetes namespace naming conventions (`.fleet.live.`, `.dev.solo.`, `.internal.`)
+- Backend service URLs that can be probed directly
+- Infrastructure geography (region names, availability zones embedded in hostnames)
+
+---
+
+## STEP 1 — Broad Internal URL Extraction (Beyond API Paths)
+
+```bash
+# Standard JS analysis only extracts API paths. This extracts ALL URLs including internal ones.
+# Download JS files first (see javascript_analysis.md STEP 1-2)
+
+# Extract ALL https:// URLs, including non-CDN internal ones
+grep -roh 'https\?://[a-zA-Z0-9._:/-]\{6,150\}' output/js_files/ 2>/dev/null \
+  | sed 's/["\`'"'"'].*$//' \
+  | sort -u \
+  | tee output/js_all_urls.txt
+
+# NOW: Split into external (known CDN/tracking) vs internal (potentially sensitive)
+grep -vE '(cdn\.|static\.|fonts\.|analytics\.|google\.|facebook\.|twitter\.|cloudflare\.|amazonaws\.com/cdn|jsdelivr\.|unpkg\.|w3\.org|schema\.org|mozilla\.|apple\.|microsoft\.|github\.com$|npmjs\.)' \
+  output/js_all_urls.txt \
+  | sort -u \
+  | tee output/js_internal_candidate_urls.txt
+
+echo "=== Internal URL candidates ==="
+cat output/js_internal_candidate_urls.txt
+echo "Total: $(wc -l < output/js_internal_candidate_urls.txt)"
+```
+
+---
+
+## STEP 2 — Fleet & Cluster Naming Convention Detection
+
+Infrastructure domains follow predictable patterns. Detect them:
+
+```bash
+# Kubernetes fleet / internal cluster patterns
+grep -iE '\.(fleet|cluster|internal|k8s|kube|svc|local|mesh|private|corp|intra)\.' \
+  output/js_internal_candidate_urls.txt | tee output/js_fleet_hostnames.txt
+
+# Environment-tagged hostnames (live/prod/staging/dev/solo/sandbox)
+grep -iE '\.(live|prod|staging|dev|sandbox|qa|uat|test|solo|preview)\.' \
+  output/js_internal_candidate_urls.txt >> output/js_fleet_hostnames.txt
+
+# Naming patterns: service-name.namespace.cluster.tld
+grep -oE '[a-z][a-z0-9-]{2,40}\.[a-z]{2,20}\.[a-z]{2,20}\.[a-z]{2,10}' \
+  output/js_internal_candidate_urls.txt \
+  | grep -vE '\.(com|net|org|io|co)\.[a-z]{2}$' \
+  >> output/js_fleet_hostnames.txt
+
+sort -u output/js_fleet_hostnames.txt | tee output/js_fleet_hostnames_dedup.txt
+echo "Fleet/cluster hostnames found: $(wc -l < output/js_fleet_hostnames_dedup.txt)"
+```
+
+---
+
+## STEP 3 — Obfuscated Hostname Detection
+
+Developers sometimes hex-encode or encode infrastructure hostnames to obscure them:
+
+```python
+# tools/decode_obfuscated_hostnames.py
+"""
+Detect and decode obfuscated hostnames in JS bundles.
+Common patterns:
+- Hex-encoded: "6f74656c" → "otel"
+- Base64-encoded: "b3RlbA==" → "otel"
+- Split strings: "ot" + "el" assembled at runtime
+- Reversed: "leto" → "otel"
+- ROT13: "bgrj" → "otel"
+"""
+import os, re, binascii, base64, codecs
+
+js_dir = "output/js_files"
+results = []
+
+for fname in os.listdir(js_dir):
+    if not fname.endswith('.js'):
+        continue
+    with open(f"{js_dir}/{fname}", 'r', errors='ignore') as f:
+        content = f.read()
+
+    # Pattern 1: Hex-encoded strings that decode to valid hostname parts
+    hex_candidates = re.findall(r'"([0-9a-f]{6,32})"', content)
+    for h in hex_candidates:
+        try:
+            decoded = binascii.unhexlify(h).decode('ascii')
+            # Only keep if decoded result looks like a hostname component
+            if re.match(r'^[a-z][a-z0-9-]{2,20}$', decoded):
+                results.append(f"HEX: {h} → {decoded}")
+        except Exception:
+            pass
+
+    # Pattern 2: Base64 strings that decode to hostnames
+    b64_candidates = re.findall(r'"([A-Za-z0-9+/]{8,50}={0,2})"', content)
+    for b in b64_candidates:
+        try:
+            decoded = base64.b64decode(b).decode('ascii')
+            if re.match(r'^[a-z][a-z0-9.-]{4,60}$', decoded):
+                results.append(f"B64: {b} → {decoded}")
+        except Exception:
+            pass
+
+    # Pattern 3: Subdomain that is hex-encoded (e.g., "6f74656c-http.target.com")
+    hex_subdomain = re.findall(r'([0-9a-f]{6,16})-[a-z]{2,10}\.[a-z0-9.-]+\.[a-z]{2,6}', content)
+    for h in hex_subdomain:
+        try:
+            decoded = binascii.unhexlify(h).decode('ascii')
+            results.append(f"HEX-SUBDOMAIN: {h} → {decoded} (likely: {decoded}-*.<domain>)")
+        except Exception:
+            pass
+
+for r in sorted(set(results)):
+    print(r)
+
+if not results:
+    print("[*] No obfuscated hostnames detected")
+```
+
+```bash
+python3 tools/decode_obfuscated_hostnames.py | tee output/js_obfuscated_hostnames.txt
+```
+
+---
+
+## STEP 4 — Probe All Internal Hostname Candidates
+
+```python
+# tools/probe_internal_hosts.py
+"""
+Probe every internal hostname candidate extracted from JS bundles.
+Test for HTTP/HTTPS accessibility on standard ports.
+Record: status code, response size, Content-Type, Server header, response body preview.
+"""
+import urllib.request, urllib.error, ssl, re, time
+
+ctx = ssl.create_default_context()
+ctx.check_hostname = False
+ctx.verify_mode = ssl.CERT_NONE
+
+results = []
+
+with open("output/js_internal_candidate_urls.txt") as f:
+    urls = [line.strip() for line in f if line.strip() and line.startswith("http")]
+
+print(f"Probing {len(urls)} internal URL candidates...")
+
+for url in urls:
+    # Normalize to just the base URL (no path) for initial probe
+    base_match = re.match(r'(https?://[a-zA-Z0-9._:-]+)', url)
+    if not base_match:
+        continue
+    base_url = base_match.group(1)
+
+    for probe_url in [url, base_url]:
+        try:
+            req = urllib.request.Request(
+                probe_url,
+                headers={"User-Agent": "Mozilla/5.0", "Accept": "application/json, text/html"}
+            )
+            with urllib.request.urlopen(req, timeout=8, context=ctx) as r:
+                body = r.read(500).decode('utf-8', 'ignore')
+                ct = r.headers.get('Content-Type', '')
+                server = r.headers.get('Server', '')
+                results.append({
+                    'status': r.status, 'url': probe_url,
+                    'ct': ct[:50], 'server': server[:30],
+                    'body': body[:200].replace('\n', ' ')
+                })
+                print(f"[{r.status}] {probe_url}")
+                print(f"  Content-Type: {ct[:60]}")
+                print(f"  Server: {server}")
+                print(f"  Body: {body[:150]}")
+                break  # Don't probe base_url if full URL worked
+        except urllib.error.HTTPError as e:
+            body = e.read(300).decode('utf-8', 'ignore')
+            results.append({'status': e.code, 'url': probe_url, 'body': body[:100]})
+            # Non-404/non-connection errors are interesting
+            if e.code not in (404, 400):
+                print(f"[{e.code}] {probe_url}: {body[:120]}")
+            break
+        except Exception as ex:
+            # Connection refused / DNS fail → not publicly accessible
+            pass
+        time.sleep(0.3)
+
+# Write results
+import json
+with open("output/js_internal_hosts_probe.txt", "w") as f:
+    for r in results:
+        f.write(json.dumps(r) + "\n")
+
+# Highlight interesting findings
+print("\n=== INTERESTING FINDINGS (non-404, public access) ===")
+for r in results:
+    if r['status'] not in (404, 400, 0) and r['status'] < 500:
+        print(f"[{r['status']}] {r['url']}: {r.get('ct','')} | {r.get('body','')[:100]}")
+```
+
+```bash
+python3 tools/probe_internal_hosts.py
+```
+
+---
+
+## STEP 5 — Analyze Error Message Patterns for Infrastructure Intelligence
+
+Error responses from internal services leak more than external services:
+
+```python
+# tools/error_intelligence.py
+"""
+Internal services often return verbose errors that reveal:
+- Framework name and version (e.g., Javalin, Spring Boot, Express)
+- Internal service names referenced in stack traces
+- Database connection strings in error bodies
+- Authentication mechanisms (JWT issuer, OAuth server URLs)
+- Internal API versioning and routing conventions
+"""
+import urllib.request, urllib.error, ssl, json
+
+ctx = ssl.create_default_context()
+ctx.check_hostname = False
+ctx.verify_mode = ssl.CERT_NONE
+
+def extract_error_intelligence(url, params_to_test=None):
+    """
+    Send malformed requests to trigger verbose errors.
+    Tests: missing params, wrong types, extra fields, empty body.
+    """
+    intelligence = {}
+
+    test_vectors = [
+        # Missing required params → often reveals param names
+        {"method": "GET", "path": url, "headers": {"Accept": "application/json"}},
+        # Wrong content type → may reveal accepted types
+        {"method": "POST", "path": url, "body": "invalid_json{{{",
+         "headers": {"Content-Type": "text/plain", "Accept": "application/json"}},
+        # Empty JSON body → may reveal required fields
+        {"method": "POST", "path": url, "body": "{}",
+         "headers": {"Content-Type": "application/json", "Accept": "application/json"}},
+        # Null bytes → often triggers framework error
+        {"method": "GET", "path": url + "?test=\x00", "headers": {}},
+    ]
+
+    for vec in test_vectors:
+        body_data = vec.get("body", "").encode() if vec.get("body") else None
+        req = urllib.request.Request(
+            vec["path"],
+            data=body_data,
+            headers=vec.get("headers", {}),
+            method=vec["method"]
+        )
+        try:
+            with urllib.request.urlopen(req, timeout=8, context=ctx) as r:
+                body = r.read(1000).decode('utf-8', 'ignore')
+                print(f"[{r.status}] {vec['method']} {url}: {body[:200]}")
+        except urllib.error.HTTPError as e:
+            body = e.read(1000).decode('utf-8', 'ignore')
+            print(f"[{e.code}] {vec['method']} {url}:")
+
+            # Parse error for intelligence
+            try:
+                err_json = json.loads(body)
+                # Javalin pattern
+                if "javalin.io" in str(err_json.get("type", "")):
+                    intelligence["framework"] = "Javalin (Kotlin/Java)"
+                    intelligence["error_message"] = err_json.get("title", "")
+                    print(f"  FRAMEWORK: Javalin detected")
+                    print(f"  ERROR: {err_json.get('title')}")
+                # Spring Boot actuator pattern
+                if "timestamp" in err_json and "path" in err_json:
+                    intelligence["framework"] = "Spring Boot"
+                # Express/Node pattern
+                if err_json.get("name") == "BadRequestError":
+                    intelligence["framework"] = "Express.js"
+            except Exception:
+                pass
+
+            # Progressive param disclosure (e.g., "Missing X parameter")
+            import re
+            missing_params = re.findall(r'[Mm]issing\s+([a-zA-Z_-]+)\s+param', body)
+            required_params = re.findall(r'[Rr]equired.*param.*?["\']([a-zA-Z_-]+)["\']', body)
+            all_params = missing_params + required_params
+            if all_params:
+                intelligence.setdefault("required_params", []).extend(all_params)
+                print(f"  PARAM DISCLOSURE: {all_params}")
+        except Exception as ex:
+            pass
+
+    return intelligence
+
+# Load probe results and analyze non-404 endpoints
+with open("output/js_internal_hosts_probe.txt") as f:
+    for line in f:
+        try:
+            r = json.loads(line)
+            if r.get("status", 404) not in (404, 0):
+                print(f"\n=== Analyzing {r['url']} ===")
+                intel = extract_error_intelligence(r['url'])
+                if intel:
+                    print(f"Intelligence: {json.dumps(intel, indent=2)}")
+        except Exception:
+            pass
+```
+
+```bash
+python3 tools/error_intelligence.py | tee output/js_error_intelligence.txt
+```
+
+---
+
+## STEP 6 — Cross-Reference Internal Hostnames with Subdomains
+
+```bash
+# Check if any internally-referenced hostnames are also public subdomains
+# This finds cases where "internal" services are accidentally internet-facing
+
+# Extract just hostnames from internal candidates
+grep -oE '[a-zA-Z0-9][a-zA-Z0-9._-]{4,80}' output/js_internal_candidate_urls.txt \
+  | grep -v '/' \
+  | sort -u > output/js_hostname_candidates.txt
+
+# Cross-reference against our subdomain list
+while IFS= read -r hostname; do
+  if grep -q "$hostname" output/subdomains.txt 2>/dev/null; then
+    echo "[IN SCOPE] $hostname is both in JS bundles AND in subdomains list"
+  fi
+done < output/js_hostname_candidates.txt | tee output/js_subdomain_crossref.txt
+
+# Also resolve any new internal hostnames that aren't in our subdomain list
+python3 -c "
+import socket, sys
+with open('output/js_hostname_candidates.txt') as f:
+    for hostname in f:
+        hostname = hostname.strip()
+        if not hostname or '.' not in hostname:
+            continue
+        try:
+            ip = socket.gethostbyname(hostname)
+            print(f'[RESOLVES] {hostname} -> {ip}')
+        except socket.gaierror:
+            pass
+" | tee output/js_new_resolving_hosts.txt
+```
+
+---
+
+## Key Patterns to Identify
+
+**Fleet/Cluster naming (indicates internal Kubernetes/container infrastructure):**
+- `.fleet.live.<company>.net` — production Kubernetes fleet
+- `.fleet.staging.<company>.net` — staging fleet
+- `<service>.dev.solo.<company>.com` — individual developer cluster
+- `<service>.<namespace>.svc.cluster.local` — internal K8s DNS (not publicly routable but leaks topology)
+- `<service>-<env>.<company>.internal` — internal DNS zone
+
+**Environment indicators in hostnames:**
+- `dev`, `staging`, `qa`, `uat`, `sandbox`, `preview` — non-production environments
+- `solo`, `personal`, `test` — individual developer environments (highest risk of being exposed)
+- `live`, `prod` — production (highest value targets)
+
+**Service type indicators:**
+- `otel`, `telemetry`, `tracing`, `metrics` — observability infrastructure
+- `billing`, `payment`, `checkout` — payment infrastructure
+- `auth`, `sso`, `login`, `oauth` — authentication infrastructure
+- `api-internal`, `internal-api`, `backend` — internal API services
+
+---
+
+## Pro Tips
+
+1. **The goldmine is in service-worker.js and webpack chunk manifests** — these load lazily and often contain backend configuration not present in the main bundle.
+2. **Check `__NEXT_DATA__` in server-rendered pages** — Next.js embeds the initial page props as JSON, which may include internal API responses with full hostnames.
+3. **Source maps (.js.map files)** — Unstrip the bundle and find commented-out internal URLs and debug configurations.
+4. **React Native / Expo bundles** — Mobile apps served from CDNs often contain more sensitive infrastructure URLs than web apps.
+5. **Service worker registration URLs** — `navigator.serviceWorker.register()` paths reveal versioned bundle URLs.
+6. **Cross-reference with historical URLs** — `output/historical_urls.txt` from Wayback Machine often has older versions of JS bundles that leaked more before the team cleaned them up.