GHSA-c2q3-p4jr-c55f
MEDIUMGHSA-c2q3-p4jr-c55f is a medium-severity (CVSS 5.4) Cross-site Scripting (XSS) vulnerability in getgrav/grav. O3 Security confirms whether GHSA-c2q3-p4jr-c55f is actually reachable in your code before you act, and blocks exploitation at runtime until you patch.
Grav Vulnerable to XSS via Taxonomy Field Values in Admin Panel
Blast Radius
getgrav/gravReal-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 Stored Cross-Site Scripting (XSS) vulnerability exists in the Grav CMS Form plugin's select field template. Taxonomy tag and category values are rendered with the Twig |raw filter in the admin panel, bypassing the global autoescape protection. An editor-level user can inject arbitrary JavaScript that executes in any administrator's browser session when they view or edit any page in the admin panel.
Additionally, Grav's built-in XSS detection (Security::detectXss()) can be bypassed by using payloads that close the <option>/<select> context and use unquoted event handlers - the on_events regex fails to match event handlers without quotes or trailing spaces before >.
Important
- The vulnerability is in the Form plugin (
select.html.twig), which is installed by default with Grav - The XSS is cross-page: a malicious taxonomy value on one page executes when an admin edits any page, because taxonomy options are rendered from a shared global pool
- An editor can exploit this without any other vulnerability - taxonomy fields are not in the server-side restricted fields list
- The
HttpOnlyflag on session cookies prevents direct session theft, but the XSS can steal the admin nonce and perform privileged actions via JavaScript
Permissions Needed
- Editor: can create or edit pages and set taxonomy tag/category values
Details
The Form plugin's select field template renders option values using the |raw Twig filter, which outputs content without HTML escaping:
File: user/plugins/form/templates/forms/fields/select/select.html.twig
{# Line 55 #}
avalue|raw
{# Line 65 #}
suboption|t|raw
{# Line 72 #}
item_value|t|raw
The taxonomy field in the page editor uses this select template. When a page has taxonomy values (tags, categories), these values are populated as <option> elements in the select dropdown. The value attribute is properly escaped by the browser's attribute encoding, but the display text between <option> tags is rendered raw:
<option value="<script>alert(1)</script>"><script>alert(1)</script></option>
Since taxonomy options are collected globally across all pages (to provide autocomplete/selection), a malicious taxonomy value on any page will appear in the taxonomy dropdown of every page editor - making this a cross-page stored XSS.
The server-side field restriction in the flex-objects plugin only blocks ['form', 'forms', 'process', 'twig'] for non-super users. Taxonomy fields are not restricted, so editors can freely set arbitrary taxonomy values.
XSS Detection Bypass
Grav's Security::detectXss() checks for dangerous_tags (e.g., <script>, <iframe>), on_events (event handlers), and invalid_protocols (e.g., javascript:). However, the on_events regex:
'on_events' => '#(<[^>]+[a-z\x00-\x20"\'\/)(?:on[a-z]+)\s*=[\s|\'"'].*[\s|\'"']>#iUu'
requires either quotes around the handler value or a trailing space before >. An unquoted handler like onerror=alert(1)> (no space before >) bypasses this check entirely.
Combined with </option></select> to break out of the select context (neither tag is in dangerous_tags), the full payload evades all three detection layers and triggers no XSS warning in the admin panel.
PoC
Step 1: Login as Editor
Navigate to http://TARGET/admin/ and authenticate with editor credentials.
Step 2: Create a Page with Malicious Taxonomy
- Go to Pages → Add → Add Page
- Title:
XSS via editor - Go to Options Tap
- On Taxonomies, Add tag:
</option></select><img src=x onerror=alert('XSS-via-editor')>
This payload:
- Closes
</option></select>to break out of the select dropdown context - Injects an
<img>tag with an unquotedonerrorhandler (bypasseson_eventsregex) - Is not in the
dangerous_tagslist (no<script>,<iframe>, etc.) - Triggers no XSS warning in the admin panel
Step 3: Trigger the XSS
When any administrator navigates to the page editor of any page (not just the malicious one), the JavaScript executes immediately.
<img width="1224" height="856" alt="image" src="https://github.com/user-attachments/assets/f008b0f2-dedb-4b22-a74a-cdc0d7325cb4" />The XSS fires because taxonomy tag options are collected globally across all pages and rendered with |raw in the select dropdown template. The payload breaks out of the <option> context, and the browser renders the <img> tag as a regular DOM element.
Impact
- Session hijacking: While
HttpOnlyprevents direct cookie theft, the XSS can steal the admin nonce token and perform any admin action via AJAX requests - Privilege escalation: An editor can perform admin-only actions (create users, modify system configuration, install plugins) through the hijacked admin session
- Cross-page impact: A single malicious taxonomy value affects the entire admin panel - every page editor view is compromised
Maintainer note — fix applied (2026-04-24)
Fixed across two repos:
-
grav-plugin-form 9.0.1 (commit
6bffb4c) — the primary fix. All four|rawfilters intemplates/forms/fields/select/select.html.twig(placeholder, avalue, suboption, item_value) have been removed. Option labels — including taxonomy values that propagate cross-page through the admin's shared selection pool — now go through Twig's default escaper, so a lower-privileged editor can no longer inject script that runs in an admin's browser when they open any page editor. -
Grav core on the
2.0branch (commit5a12f9be8, ships in 2.0.0-beta.2) — closes the detection-bypass half of the report. Theon_eventsregex inSecurity::detectXss()is tightened so unquoted handlers likeonerror=alert(1)>are flagged (see separate GHSA-9695-8fr9-hw5q), andoption/selecthave been added to defaultsecurity.xss_dangerous_tagsso</option></select>…tripwires the detector (see separate GHSA-w8cg-7jcj-4vv2).
Sites running admin2 on Grav 2.0.0-beta.2 get the 9.0.1 form plugin automatically via its existing dependency graph.
Files:
templates/forms/fields/select/select.html.twig— four|rawremoved.system/config/security.yaml— dangerous-tags list extended.system/src/Grav/Common/Security.php—on_eventsregex tightened.tests/unit/Grav/Common/Security/DetectXssTest.php— includes the GHSA-c2q3 PoC payload.
Affected Packages
| Ecosystem | Package | Vulnerable range | Fix |
|---|---|---|---|
| 🐘Packagist | getgrav/grav | all versions | 2.0.0-beta.2 |
Detection & mitigation playbook
Open-source dependencyDetect
Scan your dependency tree (package-lock.json, pnpm-lock.yaml, requirements.txt, go.sum, etc.) for getgrav/grav. 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 getgrav/grav to 2.0.0-beta.2 or later, then make sure no transitive (indirect) dependency still pins the vulnerable range — O3 confirms GHSA-c2q3-p4jr-c55f 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 GHSA-c2q3-p4jr-c55f 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 GHSA-c2q3-p4jr-c55f. Runtime protection reduces exposure until a permanent patch is applied and verified — it complements patching, it doesn't replace it.
Frequently Asked Questions
Is GHSA-c2q3-p4jr-c55f in your dependencies?
O3 detects GHSA-c2q3-p4jr-c55f across Packagist dependencies and uses function-level reachability to confirm whether the vulnerable code path is actually reachable — not just present. No false positives.