CVE-2026-92783

Yeti bridges the gap between CTI and DFIR practitioners by providing a Forensics Intelligence platform and pipeline. Prior to 2.5.1, YETI derives ExportTask.file_name from the user-controlled task name and passes it to local_storage.py without sanitizing traversal sequences or enforcing path confinement. A low-privileged authenticated user using the default local storage backend can call POST /api/v2/tasks/export/new with a traversal name and read files accessible to the YETI service account.

In Yeti prior to 2.5.1, the uuid from DFIQ YAML is not validated on import via POST /api/v2/dfiq/from_yaml and is later interpolated into an archive filesystem path in POST /api/v2/dfiq/to_archive. A low-privileged authenticated user can import a DFIQ object whose UUID contains traversal sequences and trigger archive export, writing attacker-controlled content with a .yaml suffix outside the intended temporary archive directory.

In Yeti 2.5.0 and earlier, user-controlled sorting fields and filter_aliases attribute paths are interpolated into ArangoDB AQL identifier and expression positions in core/database_arango.py. A low-privileged authenticated user can send crafted values to search endpoints, causing attacker-controlled AQL subexpressions to be parsed and evaluated.

In Yeti prior to 2.5.1, the audit router is registered without Depends(auth.get_current_active_user), and core/web/apiv2/audit.py performs no route-level authentication check for /api/v2/audit/timeline/{id}. An unauthenticated network attacker can query TimelineLog records containing analyst usernames, object identifiers and values, and before-and-after object data.