UTC
Medium

Missing authorization on gateway secret creation in MLflow basic-auth with workspaces enabled

GHSA-4449-4cjp-ffp5 · published 2026-10-28

Severity
Medium
CVSS score
4.3
Chance of exploitation (EPSS)
No CVE
Fixed in
3.17.0
What it is

Prerequisites: a valid low-privilege account on a basic-auth MLflow tracking server, multi-tenant workspaces enabled (--enable-workspaces / MLFLOW_ENABLE_WORKSPACES=true), and the AI gateway available (enabled by default in affected releases). The account may hold no workspace-wide create grant, which is exactly the state that the server rejects for CreateExperiment and CreateRegisteredModel. Demonstrated capability: the user can create arbitrary gateway provider secrets and is auto-granted MANAGE on each new secret, then read, update, or delete those specific secrets. …

What to do

Upgrade mlflow to 3.17.0 or later.

Enforce the create authorization that the design already assumes. Add a before-request validator for CreateGatewaySecret (for example validate_can_create_gateway_secret gating on _user_can_create_in_workspace, matching CreateGatewayEndpoint and CreateGatewayModelDefinition), and register it in BEFORE_REQUEST_HANDLERS. Independently, harden _authorized_outside_before_request so that mere membership in the after-request grant table no longer marks a mutating route as authorized: separate the "grant ownership from response" table from the "this route has been authorized" determination, so a create route that lacks a before-request validator still falls through to the fail-closed net. …

Affected
PackageRegistryAffectedFixed in
mlflowPyPI>= 3.10.0, < 3.17.03.17.0

The impact, patch and workaround text comes from the advisory, shortened (published by MLflow). Proof-of-concept and reproduction sections and code are left out; the original advisory has them.

Sources: GitHub Advisory Database, reviewed and malware advisories (CC BY 4.0, https://github.com/advisories); advisories published by Cursor (https://github.com/cursor/cursor/security/advisories); advisories published by OpenClaw (https://github.com/openclaw/openclaw/security/advisories); advisories published by vLLM (https://github.com/vllm-project/vllm/security/advisories); advisories published by Open WebUI (https://github.com/open-webui/open-webui/security/advisories); advisories published by n8n (https://github.com/n8n-io/n8n/security/advisories); advisories published by Flowise (https://github.com/FlowiseAI/Flowise/security/advisories); advisories published by Langflow (https://github.com/langflow-ai/langflow/security/advisories); advisories published by the MCP Python SDK project (https://github.com/modelcontextprotocol/python-sdk/security/advisories); advisories published by MLflow (https://github.com/mlflow/mlflow/security/advisories); CISA Known Exploited Vulnerabilities catalog (CC0). Data as of 2026-11-22. ShieldMap describes how each problem works and how to fix it; it does not publish exploit code. Report a correction to [email protected].