On this page (7 sections)

A critical vulnerability, tracked as CVE-2026-134292, affects SurrealDB versions from 1.0.0 up to but not including 4.0.0. According to the NVD record, the embedded JavaScript scripting engine does not enforce function capability checks when built-in functions are reached through a bridge called surrealdb.functions.

If you run SurrealDB with scripting available to authenticated users, and you rely on capability rules to restrict which built-in functions they can call, this is worth your attention today. At the time of writing, only one independent publisher covers the story, and the public record is short, so this article separates what is confirmed from what is not.

What happened

SurrealDB lets administrators control which capabilities are available to users, including which built-in functions can be called from SurrealQL. The NVD description says that a user who is denied access to a built-in function in SurrealQL can still execute it by going through surrealdb.functions from embedded JavaScript.

In short, the restriction is enforced on one path (SurrealQL) but not on another (the JavaScript bridge). The record was published on 2026-11-21. It references a GitHub security advisory from the SurrealDB project and a VulnCheck advisory. The text available to us for this article is the NVD description, so details beyond it are not confirmed here.

Technical details in plain terms

The weakness class is CWE-862, missing authorization. The system has a permission check, but one code path reaches the protected action without running it. According to the CVE description, the surrealdb.functions bridge calls two internal dispatchers, fnc::synchronous() and fnc::asynchronous(), without enforcing function capability checks.

Think of a building with a locked front door and an unlocked service entrance. The policy says certain functions are off limits, but the scripting engine's route to those functions never asks the policy. The attacker in this scenario must already be authenticated, so this is a privilege and policy bypass rather than an anonymous remote attack.

The research does not say which specific built-in functions are most dangerous to expose, what an attacker could achieve with them, or whether any default configuration is affected. Do not assume impact beyond what the CVE description states: denied built-in functions can be executed by authenticated users.

Who is realistically exposed

Exposure depends on two things: your version and how you use capabilities.

  • You run SurrealDB 1.0.0 or any later release before 4.0.0.
  • Your deployment lets authenticated users, or applications acting on their behalf, run embedded JavaScript.
  • You deny some built-in functions to those users and depend on that denial as a security control, for example in a multi-tenant setup or when giving limited access to partners or internal teams.

If nobody but fully trusted administrators can authenticate, the practical risk is lower, because they could already use those functions. If you hand out database credentials to less trusted users or services, treat the capability restriction as unreliable until you have updated.

How to check your exposure

First, find out which SurrealDB version each of your instances runs. The SurrealDB command line includes a version command; run it on the host or inside the container that runs the server and compare the result to the range 1.0.0 up to but not including 4.0.0.

Print the installed SurrealDB version (run on the server or inside the container)Shell
surreal version

If you run SurrealDB in containers, also check which image tag your deployment files reference, since a pinned older tag will keep running an affected version after you update documentation or hosts.

Next, review how your instances are started. Capability rules are normally set through startup options, so look at your service definitions, container commands or orchestration manifests and note which functions you deny and whether scripting is allowed. Consult the SurrealDB documentation for the exact option names for your version rather than guessing them.

What to do now

  1. Inventory every SurrealDB instance, including development, staging and containers, and record its version.
  2. Read the GitHub advisory for the official fix and upgrade guidance.
  3. Plan an upgrade to 4.0.0 or later, since the CVE record limits the affected range to versions before 4.0.0. Confirm the exact fixed release in the advisory, because the NVD text does not name one.
  4. Test the upgrade in staging first, as a major version change can alter behavior and configuration.
  5. Until you can upgrade, review who holds authenticated access and remove credentials that are not needed.
  6. If scripting is not required by your application, check whether you can disable it for the interim, following the official documentation for your version.
  7. Do not rely on function denial as your only barrier between untrusted users and sensitive operations. Add network restrictions so the database is reachable only from your application tier.
  8. Review database logs and application logs for unexpected function use by non-administrative accounts.
Caution. Upgrading a database server across a major version usually means a restart and may require data or configuration migration. Take a verified backup and follow the vendor's upgrade notes before touching production.

Detection and monitoring

The research does not describe any indicators of compromise or known exploitation, so there is no signature to hunt for. What you can do is watch for behavior that should not happen under your own policy.

  • Queries from restricted accounts that include embedded JavaScript when your applications never use it.
  • Use of built-in functions by users or services you have explicitly denied.
  • New or unusual authenticated sessions, especially from unexpected networks.
  • Unexpected changes to data or permissions shortly after such sessions.

If you suspect misuse, rotate the affected credentials after upgrading and review what those accounts could reach. For general guidance on responding to a possible compromise, see Website hacked: what to do.

Database servers should not be reachable from the public internet unless that is a deliberate design. An outside-in check such as a Webforta scan of a site you have verified can help you notice unexpected exposed services, though it cannot tell you whether a particular database version is vulnerable.

What is still unknown

  • The research does not state whether the flaw has been exploited in the wild, and it is not listed as a CISA KEV entry in the record provided.
  • No CVSS vector, assigner or proof-of-concept details are included, and we are deliberately not describing exploitation steps.
  • The NVD text says versions before 4.0.0 are affected but does not explicitly name a fixed release or any backported patches for older branches. Check the official advisory.

We will treat this as an evolving story. Because only one independent publisher is covered so far, rely on the vendor advisory for authoritative version and fix details.

Sources

This article is based on the public sources below. Details can change as vendors and researchers publish more; check the original advisories before acting on version numbers.

  1. CVE-2026-134292: SurrealDB 1.0.0 before 4.0.0 contains a capability check bypass in its embedded JavaScript scripting engine, where the surrealdb.functions bridge calls the…nvd.nist.gov ·
  • SurrealDB
  • CVE-2026-134292
  • database security
  • access control
  • CWE-862