CVE-2026-92684: 
JavaScript 취약점 분석 및 완화

CVE-2026-92684 is a Missing Authentication for Critical Function vulnerability (CWE-306) in the Neo4j @neo4j/graphql library that allows unauthenticated remote attackers to read protected data via GraphQL aggregation queries. It affects versions 5.0.0 through 5.12.16 and 7.0.0 through 7.6.7 of the @neo4j/graphql npm package. The vulnerability was disclosed on November 10, 2026, with patches released as versions 5.12.17 (LTS) and 7.6.8 (latest). It carries a CVSS v4.0 base score of 8.7 (High) (GitHub Advisory).

The root cause is CWE-306 (Missing Authentication for Critical Function): the @neo4j/graphql library fails to enforce field-level @authentication directives on two distinct aggregation code paths. First, string aggregate resolvers return stored field values verbatim without checking whether the requesting caller is authenticated, effectively leaking protected data. Second, connection-root aggregate queries bypass an entity's AGGREGATE authentication requirement entirely, allowing unauthenticated callers to obtain aggregate results that should be restricted. Both attack vectors require only a network-accessible GraphQL endpoint and no credentials, making exploitation straightforward for any attacker who can send GraphQL queries (GitHub Advisory).

Successful exploitation allows unauthenticated remote attackers to read data that is explicitly protected by field-level @authentication controls, undermining the confidentiality guarantees of the application's data model. String aggregate queries can return raw stored field values verbatim, potentially exposing personally identifiable information, credentials, or other sensitive graph data. Integrity and availability are not directly impacted, but the confidentiality breach could facilitate further attacks such as credential harvesting or targeted social engineering (GitHub Advisory).

There is no public proof-of-concept exploit code and no evidence of in-the-wild exploitation at the time of disclosure (GitHub Advisory). The EPSS score is reported as 0.0, and the vulnerability does not appear in the CISA Known Exploited Vulnerabilities catalog. However, the attack requires no authentication, no user interaction, and no special preconditions beyond network access to a GraphQL endpoint, making it highly automatable if targeted.

1. Reconnaissance: Identify applications exposing a Neo4j @neo4j/graphql GraphQL endpoint (versions 5.0.0–5.12.16 or 7.0.0–7.6.7) via network scanning or by inspecting npm dependency manifests in public repositories.
2. Enumerate schema: Send a GraphQL introspection query ({ __schema { types { name fields { name } } } }) to the endpoint to discover available types, fields, and aggregate query entry points.
3. Identify protected fields: Look for types with string fields that are expected to be protected (e.g., email, password hash, tokens) and locate their corresponding aggregate query names (e.g., usersAggregate).
4. Exploit string aggregate bypass: Send an unauthenticated GraphQL query targeting a string aggregate, such as:

   { usersAggregate { email { longest shortest } } }

   The library returns stored string values verbatim without enforcing @authentication, leaking protected field data.
5. Exploit connection-root aggregate bypass: Send an unauthenticated connection-root aggregate query to bypass an entity's AGGREGATE authentication requirement:

   { usersConnection { aggregate { count } } }

6. Exfiltrate data: Iterate over aggregate queries for all sensitive types to enumerate protected data at scale (GitHub Advisory).

• Network: Unusual volume of unauthenticated GraphQL POST requests to the application's GraphQL endpoint, particularly queries containing Aggregate or aggregate keywords without accompanying authentication headers or tokens.
• Logs: Application or API gateway logs showing repeated GraphQL queries of the form *Aggregate { <fieldName> { longest shortest } } or *Connection { aggregate { ... } } from unauthenticated sessions or unknown IP addresses.
• Logs: GraphQL query logs (if enabled) recording introspection queries followed immediately by aggregate queries from the same source IP, indicating schema enumeration prior to exploitation.
• Network: Outbound data transfers or API responses with unexpectedly large payloads from the GraphQL endpoint to external or unfamiliar IP addresses.

Upgrade @neo4j/graphql to version 7.6.8 (latest) or 5.12.17 (LTS) to receive the fix (GitHub Advisory). Until patching is possible, restrict access to aggregate queries over protected data at the application or API gateway level — for example, by blocking or rate-limiting GraphQL queries containing Aggregate or aggregate keywords for unauthenticated sessions. Additionally, implement supplementary authorization controls (e.g., middleware or a GraphQL firewall) to enforce authentication on all query types regardless of library-level directives.