/

CVE-2026-92804

Severity:

HIGH RISK

(7.1)

Server-Side Request Forgery (CWE-918) in Nango Platform

CVE-2026-92804

Deferred

Vulnerability type

Category:

Server-Side Request Forgery

CWE:

CWE-918

Impact:

Internal Network Access and Credential Exfiltration

Severity:

HIGH RISK

(7.1)

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

Nango through version 0.70.4 does not adequately validate caller-supplied connection configuration values used in provider token and proxy URL templates. An authenticated attacker may abuse these values to make the server send requests to internal services or cloud metadata endpoints, potentially exposing provider credentials.

Affected context

The vulnerability occurs in Nango deployments where authenticated users can influence connection configuration values that are interpolated into outbound provider token or proxy URL templates. Risk is higher when the Nango server can reach internal network services or cloud instance metadata endpoints and when outbound requests are not constrained by destination, protocol, or network policy.

Exploit preconditions

  • Attack surface: Server-side HTTP requests generated from connection configuration templates
  • Required access: Authenticated ability to create or modify relevant connection configuration values
  • Trigger point: Processing a provider token request or proxy request using the supplied configuration
  • Why exploitation is possible: Caller-controlled values can influence the destination of server-originated requests without sufficient destination validation

Why this happens

The underlying bug is insufficient validation of untrusted configuration data before it is interpolated into URLs used for server-side requests. This allows an authenticated caller to influence request destinations beyond the intended provider endpoints. The behavior maps to CWE-918, Server-Side Request Forgery, because the application acts as a network client on behalf of the caller and does not adequately restrict reachable destinations.

Potential impact

  • The server may access internal services that are not publicly reachable.
  • Cloud metadata endpoints could be queried, potentially exposing temporary credentials or instance information.
  • Provider credentials or response data may be exfiltrated.
  • Internal network reconnaissance could be performed through response behavior or request timing.
  • Downstream services could be affected if they accept the server's requests as trusted traffic.

General mitigation guidance

Mitigation generally focuses on reducing trust in caller-supplied URL components

and enforcing outbound request boundaries.

Common defensive patterns include:

  • Allowlisting approved provider hosts, schemes, ports, and URL forms.
  • Validating the final parsed destination after template interpolation rather than validating only individual fragments.
  • Blocking loopback, link-local, private, reserved, and metadata service address ranges, including equivalent resolved addresses.
  • Rechecking destinations after DNS resolution and across redirects.
  • Applying egress network policies and dedicated outbound proxies with restrictive rules.
  • Keeping provider credentials isolated from requests to untrusted destinations.
  • Running the service with least-privilege network and cloud identity permissions.
  • Treating connection configuration as untrusted input even when submitted by authenticated users.

How teams detect issues like this

Organizations can identify similar issues through:

  • Static analysis rules for HTTP clients that consume user-controlled or configuration-derived URLs.
  • Code review checks for string interpolation in token, proxy, callback, and webhook destinations.
  • Tests covering private, loopback, link-local, metadata, alternate-address, redirect, and DNS-rebinding scenarios.
  • Runtime monitoring of outbound requests to unusual hosts, internal address ranges, and cloud metadata endpoints.
  • Egress firewall or proxy policies that alert on denied or unexpected destinations.
  • Automated security review tools that trace untrusted configuration values into network request APIs.

Sources

SourceLink
github.comgithub.com/NangoHQ/nango
github.comgithub.com/NangoHQ/nango/blob/v0.70.4/packages/shared/lib/utils/utils.ts
github.comgithub.com/NangoHQ/nango/security/advisories/GHSA-hgjm-c252-ccxx
vulncheck.comvulncheck.com/advisories/nango-through-0.70.4-server-side-request-forgery-via-configuration

Frequently asked

questions

It can be exploitable through the Nango application by an authenticated user

who can influence the affected connection configuration values. The attacker

does not necessarily need direct access to the server host.

Authentication alone may not be sufficient if authorization prevents changes

to the relevant configuration. Exploitability depends on whether the caller

can submit or modify values that reach the provider token or proxy URL templates.

Treat all caller-supplied URL data as untrusted, allowlist approved destinations,

validate the fully constructed URL, and enforce network-level egress controls.

Credentials should not be available to requests sent to unapproved endpoints.

Combine secure URL construction, destination validation after resolution and

redirects, least-privilege identities, restrictive egress policies, and

regression tests for internal and metadata address ranges.

Find out what's already
exploitable in your codebase.