Description

Next AI Draw.io through 0.4.16 contains a server-side request forgery vulnerability in the POST /api/parse-url endpoint due to hostname validation that only checks string patterns without DNS resolution. Unauthenticated attackers can supply hostnames that bypass string validation but resolve to internal addresses, allowing them to reach arbitrary internal HTTP services and exfiltrate responses including cloud metadata.

Executive Summary

Technical Details

Mitigations

References

Additional Metadata

{
  "score": "8.6",
  "pubDate": "2026-08-13T19:17:32.320Z",
  "pubdate": "2026-08-13T19:17:32.320Z",
  "executiveSummary": "Next AI Draw.io through version 0.4.16 suffers from a critical Server-Side Request Forgery (SSRF) vulnerability located within the POST /api/parse-url endpoint. This flaw allows unauthenticated remote attackers to induce the server into dispatching arbitrary HTTP requests to internal resources and external destinations alike.\nThe root cause stems from insufficient hostname validation mechanisms that rely purely on static string pattern matching rather than proper DNS resolution and address sanitization. Consequently, attackers can craft malicious inputs that bypass the naive string filters while ultimately resolving to sensitive internal IP addresses or loopback interfaces.\nSuccessful exploitation of this vulnerability enables malicious actors to interact with internal services that are normally protected from the public internet. This includes accessing internal microservices, administrative panels, and cloud provider metadata endpoints.\nThe potential impact includes the unauthorized exfiltration of sensitive internal application responses, configuration data, and critical cloud infrastructure secrets such as IAM credentials. Given that the endpoint requires no authentication, the attack complexity is low, posing significant risk to confidentiality and integrity for deployments of the affected product.",
  "technicalDetails": "The vulnerability resides in the POST /api/parse-url endpoint of Next AI Draw.io through version 0.4.16. The vulnerable component is responsible for parsing and fetching data from user-supplied URLs to support drawing or diagramming functionalities.\nThe root cause of the vulnerability is flawed input validation logic. The application attempts to validate user-supplied hostnames using superficial string pattern checks (such as regular expressions or substring matching) without performing actual DNS resolution, canonicalization, or IP address range checking prior to initiating outbound network connections.\nBecause the validation operates purely at the lexical string level, attackers can construct malformed or obfuscated hostnames—such as those utilizing alternative IP representations (e.g., decimal, octal, or hexadecimal IP formats), DNS rebinding techniques, or domain aliases—that successfully pass the initial string check but resolve to internal or restricted IP addresses during the subsequent socket connection phase.\nThe attack flow proceeds as follows: An unauthenticated attacker sends a crafted HTTP POST request to the POST /api/parse-url endpoint, supplying a malicious URL parameter designed to bypass the string-based hostname filter. The server accepts the payload and passes the unvalidated destination to its internal HTTP client library. The underlying operating system resolves the hostname to an internal IP address (such as 169.254.169.254 for cloud metadata or 127.0.0.1 for local services). The server establishes an outbound TCP connection to the internal resource, executes the HTTP request, and processes the response.\nBecause the endpoint fails to adequately restrict network traffic, the application may return portions of the response or error details containing the fetched data back to the unauthenticated user. This behavior facilitates payload delivery and response exfiltration, allowing attackers to harvest cloud metadata service tokens, internal application states, and backend API responses without possessing any privileges or session tokens."
}