Mar 6, 2026·5 min read·29 visits
Mercurius versions prior to 16.8.0 fail to apply `queryDepth` limits to GraphQL subscriptions over WebSockets. Attackers can exploit this to send deeply nested queries that exhaust server resources.
A logic vulnerability in the Mercurius GraphQL adapter for Fastify allows attackers to bypass query depth limits using WebSocket subscriptions. While standard HTTP queries are validated against the configured `queryDepth`, subscription operations received via the WebSocket transport layer skip this check. This oversight allows unauthenticated remote attackers to submit arbitrarily nested queries, potentially leading to Denial of Service (DoS) via CPU and memory exhaustion when the subscription events are resolved.
Mercurius is a popular GraphQL adapter for the Fastify web framework, designed for high performance. To protect against resource exhaustion attacks—common in GraphQL due to the graph-like nature of data—Mercurius provides a queryDepth configuration option. This setting limits the maximum nesting level of incoming queries, preventing clients from requesting exponentially complex data structures.
However, a discrepancy exists in how Mercurius handles different transport protocols. While the HTTP request handler correctly invokes the depth validation logic before execution, the WebSocket handler (used for GraphQL Subscriptions) fails to enforce this check. Consequently, the queryDepth protection is effectively bypassed for any operation submitted via a persistent WebSocket connection.
This vulnerability is classified as CWE-863 (Incorrect Authorization) because the system fails to authorize the resource consumption level (depth) for a specific communication channel, despite the policy being defined globally.
The root cause lies in the architectural separation between HTTP route handling and WebSocket connection management within the Mercurius codebase. Validation logic was not centralized but rather applied independently in each transport handler.
In the HTTP path (lib/routes.js), the request processing pipeline explicitly calls the internal validation utility to check the Abstract Syntax Tree (AST) of the incoming query against the queryDepth limit. If the limit is exceeded, the request is rejected immediately with a 400 error.
In contrast, the WebSocket path is handled by lib/subscription-connection.js. The SubscriptionConnection class manages the lifecycle of WebSocket clients. When a client sends a start (or subscribe) message, the handler parses the payload and initiates the subscription iterator. Prior to version 16.8.0, this flow did not include a call to the queryDepth validator. The parameters for the depth limit were not even passed down to the subscription connection context, making validation impossible within that scope.
The fix required plumbing the queryDepthLimit configuration option down to the SubscriptionConnection class and enforcing the check during the message handling phase.
Vulnerable Code Path (Conceptual):
In lib/subscription-connection.js, the message handler simply proceeded to execution:
// Pre-patch logic
handleMessage (message) {
switch (message.type) {
case GQL_START:
// ... parsing logic ...
this.executeSubscription(id, payload, query, variables, operationName)
break;
}
}Patched Code Path:
The commit introduces the validation step explicitly. The queryDepth function is imported and called before execution proceeds.
// lib/subscription-connection.js
// [1] Import the validator
const queryDepth = require('./query-depth')
// [2] Inside the message handler
if (this.queryDepthLimit) {
// Validate the parsed document definitions
const queryDepthErrors = queryDepth(document.definitions, this.queryDepthLimit)
if (queryDepthErrors.length > 0) {
// [3] Halt execution if depth is exceeded
const err = new MER_ERR_GQL_VALIDATION()
err.errors = queryDepthErrors
this.sendMessage(GQL_ERROR, id, { payload: err })
return
}
}Additionally, index.js and lib/routes.js were updated to ensure queryDepthLimit is passed into the SubscriptionConnection constructor during initialization.
Exploiting this vulnerability requires establishing a WebSocket connection using a standard GraphQL subscription protocol (e.g., graphql-ws or subscriptions-transport-ws). No authentication is strictly required unless the endpoint itself enforces it at the connection level, but the depth check bypass occurs regardless of privilege level.
Attack Scenario:
User type with a friends field that returns a list of User types).ws://target.com/graphql.subscription {
userUpdates {
friends {
friends {
friends {
... # repeated 100+ times
name
}
}
}
}
}The primary impact is Denial of Service (DoS). While the CVSS v4.0 score is rated Low (2.7), this likely reflects the specific environmental metrics applied in the context (Availability: Low). In a real-world production environment, uncontrolled recursion can easily crash a Node.js process due to heap allocation failures or block the event loop for extended periods.
Metric Breakdown:
This vulnerability is particularly dangerous because subscriptions are long-lived. A successful attack does not require a flood of requests; a single expensive subscription can persistently drain resources every time an event is published.
CVSS:4.0/AV:N/AC:L/AT:N/PR:N/UI:N/VC:N/VI:L/VA:L/SC:N/SI:N/SA:N/E:U| Product | Affected Versions | Fixed Version |
|---|---|---|
mercurius mercurius-js | < 16.8.0 | 16.8.0 |
| Attribute | Detail |
|---|---|
| CVE ID | CVE-2026-30241 |
| CVSS v4.0 | 2.7 (Low) |
| CWE | CWE-863 (Incorrect Authorization) |
| Attack Vector | Network (WebSocket) |
| Impact | Denial of Service |
| Exploit Status | PoC Available (Regression Test) |
The software does not perform or incorrectly performs an authorization check when an actor attempts to access a resource or perform an action.
An uncontrolled resource consumption vulnerability exists in the Scala-based http4s-blaze-server package of the http4s/blaze library. The vulnerability allows remote, unauthenticated attackers to cause an Out of Memory Error (OOM) and JVM crash by streaming a continuous sequence of small or empty WebSocket continuation frames with the FIN bit set to 0. This bypasses typical payload size checks because of the JVM's per-object allocation overhead, leading to rapid heap exhaustion with minimal network bandwidth.
A critical path traversal vulnerability has been identified in the OpenList Go-based backend package. The vulnerability exists within the batch rename handler because the application does not validate the source filename parameter before constructing filesystems paths. This omission allows authenticated users to escape their designated directory and rename files in sibling paths.
OpenList version 4.2.3 and prior is vulnerable to an authorization bypass and metadata leakage. When configured with the Bleve search engine backend, OpenList fails to perform separator-aware path matching when validating tenant containment. This allows authenticated users to access sibling directories sharing similar name prefixes. Furthermore, the search backend returns unfiltered global result counts, leaking existence verification data of unauthorized files via side-channel analysis.
An authorization bypass vulnerability in OpenList version 4.2.3 and below allows authenticated users to read arbitrary files outside of their designated base directories due to an insecure path prefix check using Go's standard strings.HasPrefix function.
A security policy bypass vulnerability exists in the AWS API MCP Server (awslabs-aws-api-mcp-server) from version 0.2.13 through 1.3.46. When the server fails to load the read-only operations index during startup (due to transient network failures, file permission issues, or other exceptions), it logs a warning but continues running in an insecure, degraded state. Under this condition, the security policy engine fails open, silently skipping all subsequent security checks and consent prompts for the lifetime of the process. This permits unauthorized mutating AWS CLI commands to execute via indirect prompt injection attacks.
An incomplete escaping vulnerability in the npm package 'shescape' allows unauthenticated users to trigger dynamic shell expansions, absolute path disclosure, and command block break-outs on Unix and Windows systems.