CVE-2026-108719: LLMGateway through 1.20.0 Blind SSRF via Video-Generation callback_url

Q: Who can exploit this issue?

An attacker needs an API key and the ability to submit a video-generation request using the callback_url extension. The vulnerable webhook request originates from the worker's network, so exposure depends on what internal or cloud-metadata endpoints that worker can reach.

Q: What kinds of targets can be reached?

The supplied callback URL can point to loopback addresses, private network hosts, or cloud-metadata URLs. The worker sends deliverWebhook POST requests to those targets without the assertSafeWebhookTarget check.

Q: Are deployments affected by default?

The issue is present in LLMGateway through version 1.20.0 when API key holders can use the video-generation callback_url extension. The provided data does not state whether that extension is enabled or available by default.

Q: What can be done before a fix is available?

Restrict or disable use of the video-generation callback_url extension for API key holders. Limit the worker's network egress so it cannot reach internal services, loopback-sensitive endpoints, or cloud metadata services.

Published Oct 11, 2026

Updated

LLMGateway through 1.20.0 contains a blind server-side request forgery vulnerability that allows API key holders to reach internal hosts via the video-generation callbackurl extension. Attackers can supply loopback, private, or cloud-metadata URLs that deliverWebhook POSTs to without the assertSafeWebhookTarget check, reaching internal services from the worker's network.

Affected Software

1 affected component

LLMGateway LLMGateway<=1.20.0

Event History

Oct 11, 2026

CVE Published

via MITRE·12:19 PM

Data Sourced

via MITRE·12:19 PM

DescriptionSeverityWeakness

Data Sourced

via NVD·01:17 PM

DescriptionSeverityWeakness

Frequently Asked Questions

Who can exploit this issue?

An attacker needs an API key and the ability to submit a video-generation request using the callback_url extension. The vulnerable webhook request originates from the worker's network, so exposure depends on what internal or cloud-metadata endpoints that worker can reach.

What kinds of targets can be reached?

The supplied callback URL can point to loopback addresses, private network hosts, or cloud-metadata URLs. The worker sends deliverWebhook POST requests to those targets without the assertSafeWebhookTarget check.

Are deployments affected by default?

The issue is present in LLMGateway through version 1.20.0 when API key holders can use the video-generation callback_url extension. The provided data does not state whether that extension is enabled or available by default.

What can be done before a fix is available?

Restrict or disable use of the video-generation callback_url extension for API key holders. Limit the worker's network egress so it cannot reach internal services, loopback-sensitive endpoints, or cloud metadata services.

CVE-2026-108719: LLMGateway through 1.20.0 Blind SSRF via Video-Generation callback_url

Affected Software

Event History

Frequently Asked Questions

Who can exploit this issue?

What kinds of targets can be reached?

Are deployments affected by default?

What can be done before a fix is available?

Company

Resources

Contact

Frequently Asked Questions

Who can exploit this issue?

What kinds of targets can be reached?

Are deployments affected by default?

What can be done before a fix is available?