Summary
PrivateBin has reflected JSON injection in backend responses via unescaped REQUEST_URI
Vulnerability Details
A reflected JSON injection allows an attacker to return arbitrary data in the JSON endpoints (like /?jsonld= and /?pasteid).
Root Cause
Request::getRequestUri() sanitizes $_SERVER['REQUEST_URI'] with FILTER_SANITIZE_URL:
public function getRequestUri()
{
$uri = array_key_exists('REQUEST_URI', $_SERVER) ? filter_var($_SERVER['REQUEST_URI'], FILTER_SANITIZE_URL) : '';
return empty($uri) ? '/' : $uri;
}
FILTER_SANITIZE_URL does not strip ", ', <, > characters (per the PHP manual's allowed-character list for this filter). So the raw, attacker-controlled request URI (including query string) passes through almost unmodified into Controller::$_urlBase (set in _init()).
In Controller::_jsonld(), $_urlBase is spliced directly into one of the static .jsonld templates (js/types.jsonld, js/paste.jsonld, etc.) with a plain str_replace(), without any JSON-escaping:
$content = str_replace(
'?jsonld=',
$this->_urlBase . '?jsonld=',
file_get_contents($file)
);
...
header('Content-type: application/ld+json');
header('Access-Control-Allow-Origin: *');
header('Access-Control-Allow-Methods: GET');
echo $content;
A request URI containing a literal " therefore breaks out of the JSON string in the "@context"."pb" value and injects arbitrary attacker-controlled key/value pairs into the response body, which is served with Content-Type: application/ld+json and Access-Control-Allow-Origin: *.
Additionally, the jsonld case in Controller::__construct() returns early:
case 'jsonld':
$this->_jsonld($this->_request->getParam('jsonld'));
return;
This bypasses _setCacheHeaders() and all of the security headers normally applied in _view() (notably X-Content-Type-Options: nosniff, CSP, X-Frame-Options, Referrer-Policy). So this is the only response path lacking X-Content-Type-Options: nosniff.
Attack Scenario
- An attacker crafts a request to the target PrivateBin instance whose request-target contains a raw
"character, e.g.:GET /?jsonld=types&x="injected":"pwned","y":" HTTP/1.1
(delivered via a raw socket / HTTP client that doesn't normalize the request line, most browsers percent-encode"in the address bar, but many HTTP libraries, proxies, and automated link-preview/structured-data crawlers do not). - The server reflects the raw value into the JSON-LD response, producing a syntactically broken / attacker-extended JSON document.
- Because
Access-Control-Allow-Origin: *is set andX-Content-Type-Options: nosniffis missing on this path, any origin can fetch and rely on this manipulated content, and the response loses the defense-in-depth MIME-sniffing protection applied everywhere else in the app.
Vulnerable Code
$content = str_replace(
'?jsonld=',
$this->_urlBase . '?jsonld=',
file_get_contents($file)
);
...
header('Content-type: application/ld+json');
Verification
Dynamically confirmed on v2.0.4 (commit 597a6f0) via php -S 127.0.0.1:8082 index.php:
Request:
GET /?jsonld=types&x="injected":"pwned","y":" HTTP/1.1
Host: 127.0.0.1:8082
Connection: close
Unpatched response body (excerpt):
"pb": "/?jsonld=types&x="injected":"pwned","y":"?jsonld=types#"
i.e. the " characters are reflected raw, breaking the JSON structure, and X-Content-Type-Options is absent from the response headers.
After applying the fix above, the same request returns:
"pb": "/?jsonld=types&x=\"injected\":\"pwned\",\"y\":\"?jsonld=types#"
with X-Content-Type-Options: nosniff present, and the existing JsonApiTest::testJsonLd* unit test expectations (/?jsonld=...) remain unchanged for normal requests.
Credits
This vulnerability was reported by Iaohkut, @alanturing881, which PrivateBin would like to thank for that.
In general, PrivateBin would like to thank everyone reporting issues and potential vulnerabilities to it.
If you think you have found a vulnerability or potential security risk, we'd kindly ask you to follow our security policy and report it to us. PrivateBin then assess the report and will take the actions PrivateBin deem necessary to address it.
Impact
Reflected, unauthenticated injection of attacker-controlled content into a CORS-open application/ld+json response, plus a missing X-Content-Type-Options: nosniff header on this single response path (present everywhere else). No direct script execution was demonstrated on current browsers (this content type is generally not HTML-sniffed), but it is a real output-encoding bug (CWE-116) and a defense-in-depth gap that could be exploited by structured-data consumers or in combination with other issues / less-strict clients.
Affected versions
Security releases
Kodem intelligence
Severity tells you how bad this could be in the worst case. It does not tell you whether you are exposed. Exploitability and impact are functions of runtime truth: whether the vulnerable code is present, reachable, and actually executes in your application. A vulnerable package can sit in your dependency tree and never run.
Kodem, an Intelligent Application Security platform, uses runtime intelligence to reveal which vulnerabilities actually execute in production, so teams prioritize the ones that genuinely matter. Kodem's runtime-powered SCA identifies whether this CVE is reachable in your applications.
Already deployed Kodem?
See it in your environmentNew to Kodem? Get a demo →Remediation advice
Kodem Kai can prioritize this vulnerability in your dependency tree and generate a fix recommendation.
Frequently Asked Questions
- What is CVE-2026-55891? CVE-2026-55891 is a low-severity security vulnerability in privatebin/privatebin (composer), affecting versions <= 2.0.4. It is fixed in 2.0.5.
- Which versions of privatebin/privatebin are affected by CVE-2026-55891? privatebin/privatebin (composer) versions <= 2.0.4 is affected.
- Is there a fix for CVE-2026-55891? Yes. CVE-2026-55891 is fixed in 2.0.5. Upgrade to this version or later.
- Is CVE-2026-55891 exploitable, and should I be worried? Whether CVE-2026-55891 is exploitable in your environment depends on whether the vulnerable code is present and reachable. A CVSS score is a worst-case rating; it does not account for your specific deployment, configuration, or usage patterns. Kodem, an Intelligent Application Security platform, uses runtime intelligence to show which vulnerabilities actually execute in production, so you can focus on the ones that represent real risk. Get a demo
- What actually determines whether CVE-2026-55891 is exploitable, and how bad it is? Exploitability and impact are not fixed properties of a CVE. They depend on runtime truth: whether the vulnerable code is present, reachable, and actually executes in your application. A high CVSS score on a dependency that never runs is not the same as real risk. Kodem, an Intelligent Application Security platform, uses runtime intelligence to reveal which vulnerabilities actually execute in production, so teams prioritize the ones that genuinely matter.
- How do I fix CVE-2026-55891? Upgrade
privatebin/privatebinto 2.0.5 or later.