CVE-2026-34837: 
NixOS vulnerability analysis and mitigation

CVE-2026-34837 is a missing authorization vulnerability in Zammad, a web-based open source helpdesk/customer support system. The flaw exists in the REST endpoint POST /api/v1/ai_assistance/text_tools/:id, where context data (such as groups or organizations) supplied for use in AI prompts is not validated against the current user's access permissions. Only version 7.0.0 is affected; the issue was disclosed on April 8, 2026, and patched in version 7.0.1. It carries a CVSS v4 base score of 5.3 (Medium) (GitHub Advisory).

The root cause is classified as CWE-862 (Missing Authorization): the AI assistance controller fails to verify whether the requesting user has permission to access the context data objects (e.g., group or organization records) before incorporating them into AI prompt inputs. An authenticated attacker with any valid Zammad account can send a crafted POST request to /api/v1/ai_assistance/text_tools/:id, supplying identifiers for groups or organizations they are not authorized to view, and the system will include that data in the AI prompt response without an access check. Exploitation requires only low-level network access and a valid user session — no special privileges beyond basic authentication are needed. A technical write-up was published by the reporter (moltenbit) at https://moltenbit.net/posts/bypassing-zammad-ai-text-tool-authorization-via-rest-api (GitHub Advisory).

Successful exploitation results in unauthorized disclosure of sensitive helpdesk context data — specifically, group and organization information that the authenticated user should not have access to. The confidentiality impact is limited to the vulnerable system itself, with no integrity or availability impact reported. There is no evidence of lateral movement potential or code execution risk; the primary concern is information disclosure of organizational data through AI prompt outputs (GitHub Advisory).

There is no public proof-of-concept exploit code and no evidence of in-the-wild exploitation as of the time of disclosure. The EPSS score is approximately 0.038%, indicating a very low probability of exploitation in the near term. The vulnerability is not listed in the CISA Known Exploited Vulnerabilities (KEV) catalog. No threat actor attribution has been reported (GitHub Advisory).

1. Authentication: Log in to a Zammad instance running version 7.0.0 with any valid user account (no elevated privileges required).
2. Identify target context objects: Enumerate or guess the IDs of groups or organizations within the Zammad instance that the attacker's account is not authorized to access.
3. Craft malicious API request: Send a POST request to /api/v1/ai_assistance/text_tools/:id with a JSON body that includes the unauthorized group or organization ID as context data for the AI prompt.
4. Receive unauthorized data: The server processes the request without validating access permissions on the supplied context data, and returns AI prompt output that incorporates the restricted group or organization information, exposing it to the attacker (GitHub Advisory).

• Network: Unusual or repeated POST requests to /api/v1/ai_assistance/text_tools/ from user accounts that do not hold ticket.agent permissions, or requests referencing group/organization IDs outside the user's normal scope.
• Logs: Zammad application logs showing API calls to the AI assistance endpoint with context data parameters referencing resources not associated with the requesting user's assigned groups or organizations.
• Behavior: A single user account making high-frequency or enumeration-style requests to the AI text tools endpoint with varying :id values in a short time window.

Upgrade Zammad to version 7.0.1 or later, which contains the fix for this authorization failure. As a temporary workaround if immediate patching is not possible, restrict network-level access to the /api/v1/ai_assistance/text_tools endpoint (e.g., via firewall rules or reverse proxy ACLs) and limit AI assistance feature access to only trusted users with ticket.agent permissions (GitHub Advisory).

The vulnerability was reported by a researcher from moltenbit, who published a technical write-up at https://moltenbit.net/posts/bypassing-zammad-ai-text-tool-authorization-via-rest-api. The advisory was published by Zammad maintainer fliebe92 on GitHub. No significant broader media coverage or notable community discussion beyond standard vulnerability tracking databases has been observed (GitHub Advisory).