/

CVE-2026-72777

Severity:

HIGH RISK

(8.6)

Server-Side Request Forgery (CWE-918) in Next AI Draw.io

CVE-2026-72777

Deferred

Vulnerability type

Category:

Server-Side Request Forgery

CWE:

CWE-918

Impact:

Unauthorized access to internal services and sensitive data exposure

Severity:

HIGH RISK

(8.6)

CVSS version:

3.1

Catch this CVE while code is being written

Detect vulnerable code directly in pull requests, so CVEs are fixed before merge.

Summary

A server-side request forgery vulnerability in Next AI Draw.io allows unauthenticated users to make the application request URLs on their behalf. Because hostname checks rely on string patterns without resolving destinations, crafted hostnames may reach internal services and expose sensitive responses such as cloud metadata.

Affected context

The issue occurs in deployments where the POST /api/parse-url endpoint accepts user-controlled URLs and attempts to restrict internal destinations through hostname string validation alone. It becomes risky when the application can make outbound HTTP requests from a network position that can reach private services, link-local addresses, or cloud metadata endpoints.

Exploit preconditions

  • Attack surface: The unauthenticated POST /api/parse-url endpoint
  • Required access: Ability to submit requests to the endpoint; authentication may not be required
  • Trigger point: Server-side URL parsing or retrieval performed after accepting attacker-controlled input
  • Why exploitation is possible: Hostname pattern checks do not reliably establish the final network destination, allowing names that pass validation to resolve to internal addresses

Why this happens

The vulnerability results from treating hostname string patterns as sufficient proof that a destination is external. DNS resolution can produce an address different from what superficial validation suggests, including private, loopback, link-local, or cloud-provider metadata ranges. This maps to CWE-918: Server-Side Request Forgery, where a server is induced to send requests to attacker-selected or otherwise restricted destinations.

Potential impact

  • Internal HTTP services may be accessed without their usual network controls
  • Cloud instance metadata or temporary credentials could be exposed
  • Sensitive response contents may be returned to an external requester
  • Internal service APIs could be probed or abused
  • The application may be used as a pivot point for further attacks

General mitigation guidance

Mitigation generally focuses on reducing trust in user-provided URLs and enforcing destination policy at the network boundary.

Common defensive patterns include:

  • Prefer an allowlist of approved schemes, hosts, ports, and destinations over denylist-based hostname checks
  • Resolve hostnames and validate every resulting address before connecting
  • Re-check destination addresses across redirects and connection steps
  • Block loopback, private, link-local, multicast, and metadata address ranges
  • Restrict outbound connectivity from the application with network controls
  • Avoid returning fetched response bodies when they are not required
  • Run the URL-fetching component with minimal network and runtime privileges

How teams detect issues like this

Organizations commonly detect SSRF risks through:

  • Static analysis rules for server-side HTTP clients using user-controlled URLs
  • Code review checks for hostname-only validation, unrestricted redirects, and missing IP-range checks
  • Dynamic testing with controlled external endpoints and simulated private-address resolution
  • Egress policy monitoring for unexpected connections to internal, loopback, link-local, or metadata destinations
  • Security policies requiring URL allowlists and explicit outbound network authorization

Automated code review and application security testing tools can help identify these patterns early.

Sources

SourceLink
github.comgithub.com/DayuanJiang/next-ai-draw-io/issues/918
vulncheck.comvulncheck.com/advisories/next-ai-draw-io-ssrf-via-dns-rebinding-in-parse-url

Frequently asked

questions

Yes. The endpoint is described as unauthenticated, so a remote requester may be able to submit a URL that the server attempts to retrieve.

Access to the exposed /api/parse-url endpoint may be sufficient. The impact depends on which internal destinations the application can reach and what responses it returns.

Use an explicit destination allowlist, validate resolved IP addresses against restricted ranges, control redirects, and enforce outbound network restrictions.

Treat all user-supplied URLs as untrusted, centralize SSRF-safe fetching logic, minimize response exposure, and combine application validation with network-level egress controls.

Find out what's already
exploitable in your codebase.