Files
website/server/src/middleware/botScore.js
Claude d38c98ad9e Harden admin login: RBAC-safe controls, 2FA, bot-scoring, rate limits (#9)
Adds a layered set of protections around the admin login and the app edge.

Trust proxy (server/src/utils/trustProxy.js)
- Configurable via TRUST_PROXY; pin to the newt agent ("ptero") LAN IP so
  X-Forwarded-For is trusted ONLY from that peer. A blanket "true" is
  rejected (coerced to 1) to prevent XFF spoofing that would dodge every
  IP-based control. DEBUG_TRUST_PROXY logs peer/XFF/req.ip to re-verify the
  proxy IP without a redeploy. Documents the Omada static-reservation
  assumption.

Login throttling (server/src/middleware/loginProtection.js, rateLimit.js)
- express-slow-down progressive delay + the existing hard rate cap + a
  separate per-IP exponential backoff that persists across the rate window.
  All failures return one generic message (no user/pass disclosure).

Honeypot (login form + auth.controller)
- Hidden, plausibly-named field ("company"); a filled value fails
  generically and is scored as an unambiguous bot.

Optional per-user TOTP 2FA (speakeasy/qrcode)
- totp_secret/totp_enabled columns (+ idempotent migration). Self-service
  Account page: enroll via QR, confirm a code to enable, code-gated disable.
- Login is two-step for enrolled users: after the password, a short-lived
  signed challenge (stage:'totp', not a session) is required before the
  real session is issued.

Bot / scanner scoring + IP ban (server/src/middleware/botScore.js)
- Weighted CMS-scanner paths (this app uses none). Junk paths 404 FIRST,
  unconditionally — independent of score/ban state, so a scanner rotating
  through fresh Cloudflare IPs gets no free pass. /wp-admin/install.php is
  the top-weighted near-1-hit ban (worst offender in prod logs). Per-IP
  score with quiet-period decay temp-bans an IP from ALL routes once past a
  (deliberately low) threshold, to protect /admin from credential stuffing.
  Failed logins and honeypot hits feed the same score.
- Periodic sweep evicts stale, unbanned, quiet entries so the in-memory
  store can't grow unbounded; the interval is unref'd and cleared on
  graceful shutdown.

Tests: node --test suite (40) covering trust-proxy parsing + live req.ip
(incl. pinned-IP), rate limiter + exponential backoff, honeypot rejection,
TOTP verify (enabled/disabled) + challenge-isn't-a-session, bot-score
threshold/decay/ban + junk-404-independence + install.php + store sweep.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
2026-07-02 23:22:35 -05:00

229 lines
8.2 KiB
JavaScript

// ── Bot / scanner scoring and IP banning ──────────────────────────────────
//
// This app has no WordPress, Drupal, phpMyAdmin, .env exposure, etc. Any hit on
// those well-known scanner targets is therefore pure bot signal. Two separate
// jobs happen here, and it matters that they stay separate:
//
// 1. Junk-path 404: every hit to a known scanner path is 404'd immediately and
// UNCONDITIONALLY — independent of any IP score or ban state. Much of the
// scanning traffic here comes through Cloudflare edge ranges (104.23.x,
// 162.158.x, 172.68-71.x), i.e. a large rotating pool of source IPs, so we
// must never give a fresh IP a "free pass" on a junk path while its score
// warms up. The 404 is the primary, always-on defense.
//
// 2. Per-IP temp-ban: scoring accumulates per IP and, past a threshold, bans
// that IP from ALL routes for a while. This exists mainly to protect the
// real /admin login from credential stuffing once a scanner pivots from
// probing junk to attacking login — NOT to stop the scanning itself (fresh
// IPs are cheap for this actor, so an IP ban can't win that race). Because
// of that we bias toward a slightly LOWER threshold rather than a high one
// tuned to avoid false positives from a small/stable IP pool.
//
// State is a single-instance in-memory Map — fine for one Node process. Scores
// decay after a quiet period so a transient burst does not ban an IP forever.
//
// All time-based logic takes an optional `now` argument (defaulting to
// Date.now()) so the decay/ban windows are deterministic to test.
const log = require('../utils/logger')('botscore')
// Score at/above which an IP is banned from ALL routes. Deliberately on the low
// side (see job #2 above): fresh IPs are cheap for this actor, so we'd rather
// ban an attacking IP a little early than tune high to protect a stable pool.
const BAN_THRESHOLD = 80
// How long a ban lasts.
const BAN_MS = 60 * 60 * 1000 // 1 hour
// Quiet period after which a non-banned IP's accumulated score resets to 0.
const QUIET_MS = 30 * 60 * 1000 // 30 min
// Points added for a failed /admin login (wired in from the auth controller).
const LOGIN_FAIL_POINTS = 34
// Points for a tripped honeypot — an unambiguous bot, ban on sight.
const HONEYPOT_POINTS = BAN_THRESHOLD
// Weighted scanner paths, matched as a prefix against the lowercased request
// path, FIRST match wins — so more specific paths must precede their prefixes
// (e.g. /wp-admin/install.php before /wp-admin). Heavier weights = more damning.
//
// /wp-admin/install.php is by far the most-hit junk path in the real Pangolin
// access logs (from many rotating IPs), so it carries the single highest weight:
// a lone hit exceeds the ban threshold on its own — effectively a 1-hit ban —
// and outweighs every other individual path.
const PATH_WEIGHTS = [
['/wp-admin/install.php', 200], // top offender in prod logs — near 1-hit ban
['/.env', 100],
['/.git', 100],
['/.aws', 100],
['/wp-login.php', 100],
['/xmlrpc.php', 100],
['/wp-admin', 50],
['/administrator', 50],
['/phpmyadmin', 50],
['/mysql', 50],
['/wp-content', 40],
['/wp-includes', 40],
['/wp-json', 40],
['/user/login', 40], // Drupal
['/console', 40],
['/actuator', 40], // Spring Boot
['/vendor/phpunit', 100],
['/cgi-bin', 40],
]
// ip -> { score, lastSeen, bannedUntil }
const store = new Map()
// Return the scanner weight for a request path (0 if it is a legitimate path).
function scoreForPath(pathname) {
const p = String(pathname || '').toLowerCase()
for (const [prefix, weight] of PATH_WEIGHTS) {
if (p === prefix || p.startsWith(prefix)) return weight
}
return 0
}
function getEntry(ip) {
let e = store.get(ip)
if (!e) {
e = { score: 0, lastSeen: 0, bannedUntil: 0 }
store.set(ip, e)
}
return e
}
function isBanned(ip, now = Date.now()) {
const e = store.get(ip)
return Boolean(e && e.bannedUntil > now)
}
// Add points to an IP's score. Applies quiet-period decay first, then bans the
// IP if the new score crosses the threshold. Returns the updated entry.
function addScore(ip, points, now = Date.now(), reason = 'scan') {
const e = getEntry(ip)
// Decay: if the IP has been quiet longer than QUIET_MS (and is not currently
// banned), forget its accumulated score before adding the new hit.
if (e.bannedUntil <= now && e.lastSeen && now - e.lastSeen > QUIET_MS) {
e.score = 0
}
e.score += points
e.lastSeen = now
if (e.score >= BAN_THRESHOLD && e.bannedUntil <= now) {
e.bannedUntil = now + BAN_MS
log.warn('IP banned', { ip, score: e.score, reason, banMs: BAN_MS })
}
return e
}
// Points for a failed real login — called from the auth controller.
function recordLoginFailure(ip, now = Date.now()) {
return addScore(ip, LOGIN_FAIL_POINTS, now, 'login-fail')
}
// A tripped honeypot: instant ban-worthy score.
function recordHoneypot(ip, now = Date.now()) {
return addScore(ip, HONEYPOT_POINTS, now, 'honeypot')
}
// Early middleware: mounted before routing so banned IPs never reach a real
// handler. Everything here 404s (never 403) so we never confirm a path or a ban.
function guard(req, res, next) {
const ip = req.ip
const now = Date.now()
// (1) Known junk/scanner path → 404 FIRST, unconditionally. This is evaluated
// and returned before any ban check, so the 404 is fully independent of this
// IP's score/ban state: a scanner cycling through fresh Cloudflare IPs gets no
// free pass on a junk path. Scoring still runs (it accrues toward a /admin ban
// if the IP is reused), but the 404 does not depend on it.
const points = scoreForPath(req.path)
if (points > 0) {
const e = addScore(ip, points, now, 'scan')
log.warn('scanner path hit', { ip, path: req.path, points, score: e.score })
return notFound(res)
}
// (2) Non-junk path → block only if this IP is already banned (the
// credential-stuffing guard for the real /admin login), else let it through.
if (isBanned(ip, now)) {
log.debug('blocked banned IP', { ip, path: req.path })
return notFound(res)
}
return next()
}
// Uniform 404 — mirrors the SPA/API "Not found" shape without leaking anything.
function notFound(res) {
return res.status(404).json({ message: 'Not found' })
}
// ── Cleanup sweep ──────────────────────────────────────────────────────────
// Every unique IP that hits a scored path adds an entry and nothing else evicts
// it, so the store would grow unbounded. Periodically drop entries that are no
// longer meaningful: NOT banned and quiet longer than QUIET_MS (their score
// would already reset to 0 on next touch anyway). Banned entries, and entries
// still inside their quiet decay window, are left untouched. Returns the count
// removed. The eviction age reuses QUIET_MS; SWEEP_INTERVAL_MS is only cadence.
const SWEEP_INTERVAL_MS = 10 * 60 * 1000 // 10 min
function sweep(now = Date.now()) {
let removed = 0
for (const [ip, e] of store) {
if (e.bannedUntil <= now && now - e.lastSeen > QUIET_MS) {
store.delete(ip)
removed++
}
}
if (removed > 0) log.debug('store sweep', { removed, remaining: store.size })
return removed
}
let sweepTimer = null
function startSweeper() {
if (sweepTimer) return sweepTimer
sweepTimer = setInterval(() => sweep(), SWEEP_INTERVAL_MS)
// Never let the sweep timer alone keep the event loop alive (tests, shutdown).
if (sweepTimer.unref) sweepTimer.unref()
return sweepTimer
}
function stopSweeper() {
if (sweepTimer) {
clearInterval(sweepTimer)
sweepTimer = null
}
}
// Start sweeping on load — this is a single long-lived process.
startSweeper()
// Test/ops helpers.
function _reset() {
store.clear()
}
function _snapshot(ip) {
const e = store.get(ip)
return e ? { ...e } : null
}
module.exports = {
guard,
scoreForPath,
addScore,
isBanned,
recordLoginFailure,
recordHoneypot,
sweep,
startSweeper,
stopSweeper,
_reset,
_snapshot,
// Exported for tests / tuning.
BAN_THRESHOLD,
BAN_MS,
QUIET_MS,
LOGIN_FAIL_POINTS,
HONEYPOT_POINTS,
SWEEP_INTERVAL_MS,
}