GHSA-HMFR-RX46-4JX2

Vulnerability from github – Published: 2025-08-26 18:42 – Updated: 2025-08-26 18:42
VLAI
Summary
GraphQL Armor Max-Depth Plugin Bypass via Introspection Query Obfuscation
Details

Summary

A query depth restriction using the max-depth property can be bypassed if ignoreIntrospection is enabled (which is the default configuration) by naming your query/fragment __schema.

Details

At the start of the countDepth function, we have the following check for the ignoreIntrospection option:

    if (this.config.ignoreIntrospection && 'name' in node && node.name?.value === '__schema') {
        return 0;
    }

However, the node can be one of: FieldNode, FragmentDefinitionNode, InlineFragmentNode, OperationDefinitionNode, FragmentSpreadNode.

For example, consider sending the following query:

query hello {
  books {
    title
  }
}

This would create an OperationDefinitionNode where node.name.value == 'hello'

The proper way to handle this is to check explicitly for the __schema field, which corresponds to a FieldNode.

The fix is

    if (
      this.config.ignoreIntrospection &&
      'name' in node &&
      node.name?.value === '__schema' &&
      node.kind === Kind.FIELD
    ) {
      return 0;
    }

This ensures that the node is explicitly a FieldNode.

PoC

Max depth: 6

query {
  books {
    author {
      books {
        author {
          ...__schema
        }
      }
    }
  }
}
fragment __schema on Author {
  books {
    title
  }
}

Impact

This issue affects applications using the GraphQL Armor Depth Limit plugin with ignoreIntrospection enabled.

Fix

This is fixed in PR#823

Severity
5.3 (Medium)
CVSS:3.0/AV:N/AC:L/PR:N/UI:N/S:U/C:N/I:N/A:L
Show details on source website
To clipboard 

{
  "affected": [
    {
      "database_specific": {
        "last_known_affected_version_range": "\u003c= 2.4.1"
      },
      "package": {
        "ecosystem": "npm",
        "name": "@escape.tech/graphql-armor-max-depth"
      },
      "ranges": [
        {
          "events": [
            {
              "introduced": "0"
            },
            {
              "fixed": "2.4.2"
            }
          ],
          "type": "ECOSYSTEM"
        }
      ]
    }
  ],
  "aliases": [],
  "database_specific": {
    "cwe_ids": [
      "CWE-400"
    ],
    "github_reviewed": true,
    "github_reviewed_at": "2025-08-26T18:42:37Z",
    "nvd_published_at": null,
    "severity": "MODERATE"
  },
  "details": "### Summary\nA query depth restriction using the `max-depth` property can be bypassed if `ignoreIntrospection` is enabled (which is the default configuration) by naming your query/fragment `__schema`.\n\n### Details\nAt the start of the `countDepth` function, we have the following check for the `ignoreIntrospection` option:\n\n```typescript\n    if (this.config.ignoreIntrospection \u0026\u0026 \u0027name\u0027 in node \u0026\u0026 node.name?.value === \u0027__schema\u0027) {\n        return 0;\n    }\n```\n\nHowever, the `node` can be one of: `FieldNode`, `FragmentDefinitionNode`, `InlineFragmentNode`, `OperationDefinitionNode`, `FragmentSpreadNode`.\n\nFor example, consider sending the following query:\n\n```graphql\nquery hello {\n  books {\n    title\n  }\n}\n```\n\nThis would create an `OperationDefinitionNode` where `node.name.value == \u0027hello\u0027`\n\nThe proper way to handle this is to check explicitly for the `__schema` field, which corresponds to a `FieldNode`.\n\nThe fix is\n\n```typescript\n    if (\n      this.config.ignoreIntrospection \u0026\u0026\n      \u0027name\u0027 in node \u0026\u0026\n      node.name?.value === \u0027__schema\u0027 \u0026\u0026\n      node.kind === Kind.FIELD\n    ) {\n      return 0;\n    }\n```\n\nThis ensures that the node is explicitly a `FieldNode`.\n\n### PoC\n\nMax depth: `6`\n\n```graphql\nquery {\n  books {\n    author {\n      books {\n        author {\n          ...__schema\n        }\n      }\n    }\n  }\n}\nfragment __schema on Author {\n  books {\n    title\n  }\n}\n```\n\n### Impact\n\nThis issue affects applications using the GraphQL Armor Depth Limit plugin with `ignoreIntrospection` enabled.\n\n### Fix\n\nThis is fixed in [PR#823](https://github.com/Escape-Technologies/graphql-armor/pull/823)",
  "id": "GHSA-hmfr-rx46-4jx2",
  "modified": "2025-08-26T18:42:37Z",
  "published": "2025-08-26T18:42:37Z",
  "references": [
    {
      "type": "WEB",
      "url": "https://github.com/Escape-Technologies/graphql-armor/security/advisories/GHSA-hmfr-rx46-4jx2"
    },
    {
      "type": "WEB",
      "url": "https://github.com/Escape-Technologies/graphql-armor/pull/823"
    },
    {
      "type": "WEB",
      "url": "https://github.com/Escape-Technologies/graphql-armor/commit/1f923bc09f5f053f60b6ba2bd419d4b94cbe1db3"
    },
    {
      "type": "PACKAGE",
      "url": "https://github.com/Escape-Technologies/graphql-armor"
    }
  ],
  "schema_version": "1.4.0",
  "severity": [
    {
      "score": "CVSS:3.0/AV:N/AC:L/PR:N/UI:N/S:U/C:N/I:N/A:L",
      "type": "CVSS_V3"
    }
  ],
  "summary": "GraphQL Armor Max-Depth Plugin Bypass via Introspection Query Obfuscation"
}



Log in or create an account to share your comment.




Tags
Taxonomy of the tags.





Forecast uses a logistic model when the trend is rising, or an exponential decay model when the trend is falling. Fitted via linearized least squares.

Sightings

Author Source Type Date Other

Nomenclature

  • Seen: The vulnerability was mentioned, discussed, or observed by the user.
  • Confirmed: The vulnerability has been validated from an analyst's perspective.
  • Published Proof of Concept: A public proof of concept is available for this vulnerability.
  • Exploited: The vulnerability was observed as exploited by the user who reported the sighting.
  • Patched: The vulnerability was observed as successfully patched by the user who reported the sighting.
  • Not exploited: The vulnerability was not observed as exploited by the user who reported the sighting.
  • Not confirmed: The user expressed doubt about the validity of the vulnerability.
  • Not patched: The vulnerability was not observed as successfully patched by the user who reported the sighting.



Related by attack behaviour

Vulnerabilities whose description is nearest to this one in the vector space of the CIRCL/vulnerability-attack-technique-biencoder model. This is a similarity search over the bi-encoder space (plain cosine), not a classification, and it has no measured accuracy.