visualTicket: 14 CVEs in a Ticketing Platform
CERT Polska disclosed 14 visualTicket flaws spanning IDOR, path traversal, XSS, CSRF and data exposure. Here is a technical remediation plan.
- AUTHOR
- Karol Rapacz / CEO Breachroad · OSCP · PNPT
- PUBLISHED
- 16 November 2026
- READING TIME
- 22 min read
- TOPIC
- Vulnerabilities and CVEs
On 16 November 2026, CERT Polska published a coordinated disclosure covering 14 vulnerabilities in VisualNet visualTicket. The findings include broken object-level authorisation, path traversal, session fixation, XSS, CSRF, JSONP data exposure, unprotected invoice files, detailed error leakage and disclosure of a Polish KSeF integration secret.
The case demonstrates why “one update” may remove an attack chain rather than fourteen unrelated defects. Some issues have limited impact in isolation. In combination, they can help an attacker learn the application’s structure, control a session, execute an action in a user’s context, modify a cart or reach confidential documents.
According to the English-language CERT Polska advisory, all visualTicket versions older than 20260817 are affected. The issues were fixed or partially mitigated in that release. CERT Polska coordinated the disclosure after a report by Aleksander Hubich.
Who needs to investigate
visualTicket supports ticket sales and reservation workflows. An organisation should not search only for the product name in a licence inventory. Check public sales domains, application servers, deployment repositories, DNS records, supplier contracts and panels used by a cinema, cultural institution, venue or event operator.
The initial questions are:
- do we run visualTicket, and which version is actually deployed;
- is the application reachable from the Internet;
- is the administrative panel exposed publicly;
- does the live artefact match the version recorded in documentation;
- are custom modifications blocking a straightforward upgrade;
- what information is present in invoices, carts, reports and KSeF integration;
- who held Author and administrator roles;
- how long are HTTP, application and administrative logs retained?
Do not assume a footer version is authoritative. Confirm the release from the artefact, vendor panel or deployment digest.
Mapping the fourteen vulnerabilities
CERT Polska assigned identifiers CVE-2026-113443 through CVE-2026-113456. Defenders can make the set easier to reason about by grouping the findings around the failed control.
Authorisation and data isolation
CVE-2026-113443 is an IDOR in the cart service. The server trusted a supplied cart_id without verifying that the cart belonged to the current session. Sequential identifiers made valid values easier to predict. An attacker with an independent session could change or remove another user’s cart items and affect a reservation’s total.
CVE-2026-113453 affected PDF invoices exposed at a static web path without session checks. The invoice register required authentication, but the file itself did not enforce the same decision. It is a classic case of protecting the interface rather than the underlying resource.
CVE-2026-113449 exposed session and cart data through JSONP endpoints when an attacker possessed the session token. Release 20260817 introduced a SameSite=Lax mitigation. That setting limits some cross-site requests, but it does not replace object-level authorisation.
Files and path handling
CVE-2026-113444 is a path-traversal issue in thumbnail generation. The file parameter was used to construct an output path without effectively confining it to the uploads directory. CERT describes the ability to create JPEG files outside the intended directory, including writable locations under the web root.
Precision matters: the advisory does not describe arbitrary-content arbitrary-file write or confirmed remote code execution. The stated impact is unauthorised JPEG creation. Any further consequence depends on server configuration, writable paths and file handling.
Sessions and browser-side execution
CVE-2026-113447 concerns session fixation. The application accepted a Symfony session identifier in the URL and set it as a cookie. An attacker could cause a victim to use a known identifier and later reuse the session after authentication.
Several XSS findings affected language and culture parameters, the login function, a JSONP callback and the selectedDate value on an order summary. They are CVE-2026-113446, CVE-2026-113448, CVE-2026-113450 and CVE-2026-113452. XSS executes in the application’s origin, so practical impact depends on reachable data, Content Security Policy, cookie attributes and browser-accessible functions.
CSRF and state-changing operations
CVE-2026-113451 covered several state-changing endpoints available through GET without CSRF protection. A hostile page could induce the browser of a logged-in user, including an administrator, to perform an unintended operation. In version 20260817, settings changes and bulk refunds require POST plus a CSRF token, and the session cookie uses SameSite=Lax.
CERT describes the remediation as partial. The system owner should therefore inventory every state-changing action, not just those named in the advisory. POST alone is not CSRF protection; the application also needs an unpredictable session-bound token and appropriate origin validation.
Diagnostic and secret exposure
CVE-2026-113445 exposed SQL/ORM details when an invalid payment_type was supplied during checkout. CVE-2026-113454 and CVE-2026-113456 involved detailed exceptions and SQL statements returned by administrative and cinema report generators.
CVE-2026-113455 displayed the real KSeF API token in the administrator settings panel without masking it. Masking cannot replace access control, but it reduces accidental exposure through screen sharing, recordings, screenshots or access by a person who may change configuration without needing to retrieve the current secret.
The priority upgrade: version 20260817
The first action is to confirm and plan an upgrade to 20260817 or a later vendor-supported release. Organisations running a customised build should not copy individual fixes by sight. They need a controlled backup, test environment, database-schema comparison, end-to-end purchase test and rollback plan.
Record the version, time, owner and artefact digests before deployment. Afterwards, confirm that every cluster node and disaster-recovery copy has been updated. One legacy server behind a load balancer can continue serving vulnerable traffic.
If an immediate upgrade is impossible, restrict public panel access, remove unnecessary accounts and roles, reduce the web process’s filesystem rights, increase monitoring and consider temporarily disabling the highest-risk functions. A WAF may obstruct some XSS or path-traversal attempts, but it cannot fix IDOR and is not a complete remedy.
Prioritise exposure rather than adding CVSS scores
Fourteen CVEs do not represent fourteen equally likely incidents. Priority depends on exposure, required privileges, the reachable asset and business impact. An Internet-facing ticketing system with an administrative panel and invoice data has a different risk profile from an isolated test instance containing synthetic data.
Address paths first when they:
- are available without authentication;
- affect sessions or authorisation;
- lead to financial documents or personal data;
- allow action in an administrator’s context;
- change sales, refunds or configuration;
- can be combined into a coherent chain.
Our guide to prioritising critical vulnerabilities explains this approach. The number of CVEs is a signal, not a remediation queue.
A safe compromise review
An upgrade does not answer whether a flaw was exploited beforehand. Preserve reverse-proxy, web-server, application, database, filesystem and administrative logs before retention removes them. Record timestamps and time-zone configuration.
Investigators can look for:
- unusual or sequential cart_id values referenced across separate sessions;
- encoded path separators and attempted directory traversal;
- unexpected JPEG files outside the upload directory;
- session identifiers passed in URLs;
- unusual JSONP callbacks and XSS-like parameters;
- administrative actions immediately following navigation from an external site;
- invoice downloads without preceding authorised navigation;
- repeated SQL/ORM errors from a common source;
- reads or changes to KSeF settings;
- bulk refunds, configuration changes and cart anomalies.
A single log entry is not proof of compromise. Search engines, quality scanners, authorised testers and broken clients can also generate unusual requests. Correlate the request with its session, account, response, state change and database trace.
Verify the fix without publishing an exploit
A retest should confirm the security property rather than merely the absence of one known payload. For IDOR, verify that the server always binds a cart to the current session. For invoices, enforce authorisation on the file resource. For path traversal, validate canonicalisation and confinement to an approved directory.
For session fixation, expect an identifier to be regenerated after login and reject any session supplied through a URL. For XSS, verify context-sensitive output encoding rather than a character blacklist. For CSRF, confirm token handling, SameSite attributes and correct HTTP methods. For the KSeF secret, ensure the interface cannot retrieve the stored value and only supports replacing it.
This work belongs in a web application penetration test or a broader web and API penetration test. Testing requires explicit authorisation and a controlled environment; it must never access another customer’s data.
The software-design lesson
The recurring pattern is that a security decision was enforced somewhere other than the resource that required protection. The UI hides an invoice while the file remains public. A cart has an identifier, but the server does not check ownership. A panel requires authentication, but an action accepts GET without CSRF protection. Only administrators see a secret, although not every administrator needs to read it.
A secure design centralises authorisation and applies it to every object. An identifier is not a permission. State-changing actions require explicit intent. Client errors should be generic while protected logs retain details. Secrets need to be replaceable and rotatable but unreadable after storage.
These principles map directly to OWASP API Security Top 10 risks and the object-authorisation tests in our IDOR and BOLA testing guide.
Actions for 24 hours, seven days and 30 days
Within 24 hours, identify the owner, version, exposure and upgrade path. Preserve logs, restrict the panel, review privileged accounts and decide whether the KSeF token requires rotation. Do not rotate a secret before cutting off active access and preserving evidence.
Within seven days, update every instance, run sales-regression tests and retest the security controls. Review telemetry for the vulnerability classes described above. If evidence of data access appears, start formal incident response and assess legal notification duties.
Within 30 days, remove the systemic causes: fragmented authorisation, broad roles, public administration, excessive web-process rights, readable secrets and incomplete logging. Add negative tests to CI/CD so the same bug classes do not return in a later release. A mature DevSecOps security programme supports that control.
Source facts and Breachroad conclusions
CERT Polska confirms the CVE identifiers, affected-version boundary, vulnerability mechanics and remediation state in release 20260817. The advisory does not report active exploitation in customer environments. That omission does not prove no attempt ever occurred, but exploitation must not be presented as confirmed.
The priority model, detection patterns, 24/7/30 plan and architectural recommendations are Breachroad analysis. Their purpose is to turn disclosure into safe action, not to publish an attack recipe.
Summary
visualTicket owners should verify the deployed version, move to 20260817 or later, preserve logs and investigate possible abuse of sessions, carts, invoices, reports and KSeF settings. Updating is the first step, but it does not replace a retrospective evidence review.
The larger lesson applies to every application: access control must protect each object and operation on the server. A hidden link, difficult identifier, POST request or administrator panel is not a security boundary on its own.


