CVE-2026-82638: jina-ai reader Server-Side Request Forgery via disabled private-address guard
jina-ai reader disables its private-address guard outside Google Cloud deployments, allowing unauthenticated attackers to perform server-side request forgery. Attackers can supply publicly resolvable hostnames mapping to private addresses to retrieve cloud metadata and internal service content.
Affected Software
Event History
Frequently Asked Questions
Which deployments are exposed?
Deployments of jina-ai reader outside Google Cloud are exposed because the private-address guard is disabled in those environments. Google Cloud deployments are not identified as affected by this condition.
What does an attacker need to exploit this issue?
An attacker needs only network access to the affected reader service and a publicly resolvable hostname that maps to a private address. No authentication or user interaction is required.
What could an attacker access through the SSRF?
The issue can be used to retrieve cloud metadata and content from internal services reachable by the reader server. The reported impact is confidentiality exposure; integrity and availability impacts are not indicated.
What configuration should be prioritized if patching is not immediately possible?
Ensure the service does not resolve or connect to private addresses through attacker-supplied hostnames, since the disabled private-address guard is the stated cause. Restrict the reader service's network access to cloud metadata endpoints and internal services where possible.