Skip to content

NVIDIA Triton Model Repository Control Exposed

triton-model-control-exposed is a high severity check in the ai category, written in yaml. Its source is templates/ai/triton/triton-model-control-exposed.yaml in cert-x-gen-templates.

Detects an exposed NVIDIA Triton model repository control endpoint (/v2/repository/index). When model-control is enabled without proper access controls, attackers may load or modify models and potentially abuse path traversal issues (CVE-2023-31036).

Field Value
Id triton-model-control-exposed
Severity high
Language yaml
Category ai
Author CERT-X-GEN Security Team
Template version 1.0.0
Confidence 80
CVSS 8.6
Weakness CWE-22, CWE-306
Tags triton, nvidia, model-control, repository, path-traversal, unauth, cve-2023-31036
Target kind not recorded
Oracle not recorded

The file declares cve_ids, http, remediation as well. Those are not tabled above; read the source for what they carry.

Declared in the header. cxg parses @references and then discards it, and nothing at scan time reads it, so this is the only place the links a template cites are surfaced.

To see what the copy on your machine says about itself, and to confirm it is installed at all:

Terminal window
cxg template info triton-model-control-exposed

The id it prints is the one to pass anywhere a template is selected. See cxg template for the rest of the subcommand, Scan a target for running a scan, and A match is not a finding for how to read what comes back.