CVE-2026-92566: DataGear SSRF Vulnerability

CVE-2026-92566 Overview

CVE-2026-92566 is an unauthenticated Server-Side Request Forgery (SSRF) vulnerability in DataGear through version 6.0.0. The flaw resides in the /dataSet/preview/Http endpoint, which accepts caller-controlled URIs without validation or authentication. Attackers can force the DataGear server to issue arbitrary GET, POST, PUT, PATCH, or DELETE requests to internal endpoints and cloud metadata services. Full response bodies are returned to the attacker, enabling reconnaissance of internal networks and theft of cloud instance credentials. The vulnerability is classified under CWE-918: Server-Side Request Forgery.

Critical Impact
Unauthenticated attackers can pivot into internal networks, access cloud metadata endpoints such as 169.254.169.254, and retrieve IAM credentials from the DataGear host.

Affected Products

DataGear versions up to and including 6.0.0
The HttpDataSet component in datagear-analysis
Web configuration exposed via CoreConfigSupport and SecurityConfigSupport in datagear-web

Discovery Timeline

2026-09-16 - CVE-2026-92566 published to NVD
2026-09-16 - Last updated in NVD database

Technical Details for CVE-2026-92566

Vulnerability Analysis

DataGear exposes the /dataSet/preview/Http endpoint to preview data returned from HTTP-based datasets. The endpoint accepts a URI parameter and instantiates an HttpDataSet that issues the outbound HTTP request from the server. No allowlist, authentication check, or network egress control is applied before the request is executed. The endpoint is reachable without a session because the security configuration in SecurityConfigSupport permits anonymous access to dataset preview routes. The server returns the raw response body, including headers and payload, to the caller.

Root Cause

The root cause is a combination of two design flaws. First, HttpDataSet treats the user-supplied URI as trusted input and forwards it directly to the underlying HTTP client. Second, SecurityConfigSupport places the preview endpoint outside the authenticated request matchers. Together, these choices allow any unauthenticated caller to weaponize the DataGear backend as an HTTP proxy that speaks all common verbs.

Attack Vector

An attacker sends a crafted request to /dataSet/preview/Http specifying a target URI and HTTP method. Because DataGear performs the request server-side, the attacker can reach loopback services, internal-only management interfaces, Kubernetes API servers, and cloud provider metadata endpoints. The full response is echoed back, which enables extraction of instance role credentials, service tokens, and internal configuration data. See the VulnCheck Security Advisory for DataGear for the technical write-up.

The vulnerable request handling is described in the HttpDataSet source, while the missing authentication requirement is visible in the SecurityConfigSupport source.

Detection Methods for CVE-2026-92566

Indicators of Compromise

Unauthenticated HTTP requests to /dataSet/preview/Http in DataGear access logs
Outbound requests from the DataGear host to 169.254.169.254, metadata.google.internal, or other cloud metadata endpoints
Server-initiated requests to RFC1918 addresses that do not match normal DataGear dataset sources
HTTP methods such as PUT, PATCH, or DELETE originating from the DataGear process to internal services

Detection Strategies

Inspect reverse proxy and application logs for POST bodies targeting /dataSet/preview/Http with URI parameters pointing to internal ranges
Correlate DataGear process network connections with destination IPs outside the expected dataset provider set
Alert on any DataGear-originated request to link-local addresses (169.254.0.0/16)

Monitoring Recommendations

Enable egress logging on hosts running DataGear and forward events to a centralized analytics tier
Baseline expected outbound destinations from the DataGear service account and alert on deviations
Monitor cloud provider audit logs for metadata service token issuance tied to the DataGear instance role

How to Mitigate CVE-2026-92566

Immediate Actions Required

Restrict network access to the DataGear web interface using perimeter controls and internal firewalls
Block outbound traffic from the DataGear host to cloud metadata endpoints and RFC1918 ranges that are not required
Require authentication in front of DataGear by placing it behind a reverse proxy that enforces access control on /dataSet/preview/*
Rotate any IAM credentials, tokens, or secrets accessible from the DataGear host if exploitation is suspected

Patch Information

No vendor patch is referenced in the NVD record at time of publication. Track GitHub Issue #37 and the DataGear repository for a fixed release. Apply the update as soon as one is published by the maintainers.

Workarounds

Deploy DataGear on a host with IMDSv2 required and hop-limit set to 1 to break metadata SSRF chains on AWS
Front the application with a web application firewall rule that blocks requests to /dataSet/preview/Http from untrusted networks
Run DataGear inside a network namespace or container with an egress policy that only permits approved dataset endpoints

bash

# Example nginx rule to block unauthenticated access to the vulnerable endpoint
location ~ ^/dataSet/preview/Http {
    allow 10.0.0.0/8;
    deny all;
    auth_basic "DataGear";
    auth_basic_user_file /etc/nginx/.htpasswd;
    proxy_pass http://datagear_backend;
}