Amazon Data Processing MCP Server

The AWS DataProcessing MCP server provides AI code assistants with comprehensive data processing tools and real-time pipeline visibility across AWS Glue and Amazon EMR-EC2. This integration equips large language models (LLMs) with essential data engineering capabilities and contextual awareness, enabling AI code assistants to streamline data processing workflows through intelligent guidance — from initial data discovery and cataloging through complex ETL pipeline orchestration and big data analytics optimization.

Integrating the DataProcessing MCP server into AI code assistants transforms data engineering workflows across all phases, from simplifying data catalog management with automated schema discovery and data quality validation. Additionally, it streamlines ETL job creation with intelligent code generation and best practice recommendations. It accelerates big data processing through automated EMR cluster provisioning and workload optimization. Finally, it enhances troubleshooting through intelligent debugging tools and operational insights. All of this simplifies complex data operations through natural language interactions in AI code assistants.

Key features

AWS Glue Integration

• Data Catalog Management: Enables users to explore, create, and manage databases, tables, and partitions through natural language requests, automatically translating them into appropriate AWS Glue Data Catalog operations.
• Connection Type Discovery: Discover and describe available AWS Glue connection types, including their supported properties, authentication methods, and compute environments.
• Connection Metadata & Entity Exploration: Explore entities available through Glue connections (e.g., SaaS objects, database tables), describe entity schemas, and preview entity data records from connected data sources.
• Interactive Sessions: Provides interactive development environment for Spark and Ray workloads, enabling data exploration, debugging, and iterative development through managed Jupyter-like sessions.
• Workflows and Triggers: Orchestrates complex ETL activities through visual workflows and automated triggers, supporting scheduled, conditional, and event-based execution patterns.
• Commons: Enables users to create and manage usage profiles, security configurations, catalog encryption settings and resource policies, which provide users with the ability to manage the configuration and encryption of several Glue resources like ETL jobs, catalogs, etc.
• ETL Job Orchestration: Provides the ability to create, monitor, and manage Glue ETL jobs with automatic script generation, job scheduling, and workflow coordination based on user-defined data transformation requirements.
• Crawler Management: Enables intelligent data discovery through automated crawler configuration, scheduling, and metadata extraction from various data sources.

Amazon EMR Integration

• Cluster Management: Enables users to create, configure, monitor, and terminate EMR clusters with comprehensive control over instance types, applications, and configurations through natural language requests.
• Instance Management: Provides the ability to add, modify, and monitor instance fleets and instance groups within EMR clusters, supporting both on-demand and spot instances with auto-scaling capabilities.
• Step Execution: Orchestrates data processing workflows through EMR steps, allowing users to submit, monitor, and manage Hadoop, Spark, and other application jobs on running clusters.
• Security Configuration: Manages EMR security settings including encryption, authentication, and authorization policies to ensure secure data processing environments.

Amazon Athena Integration

• Query Execution: Enables users to execute, monitor, and manage SQL queries with comprehensive control over query lifecycle, including starting queries, retrieving results, monitoring performance statistics, and canceling running queries through natural language requests.
• Named Query Management: Provides the ability to create, update, retrieve, and delete saved SQL queries, enabling users to build reusable query libraries with proper organization and team collaboration capabilities.
• Data Catalog Operations: Manages Athena data catalogs with support for multiple catalog types (LAMBDA, GLUE, HIVE, FEDERATED), enabling users to create, configure, and maintain data source connections for cross-platform querying.
• Database and Table Discovery: Facilitates data exploration through comprehensive database and table metadata retrieval, allowing users to discover available data sources, understand schema structures, and navigate data catalogs efficiently.
• Workgroup Administration: Orchestrates query execution environments through workgroup management, providing cost control, access management, and query result configuration with support for different user groups and organizational policies.

Prerequisites

• Install Python 3.10+
• Install the uv package manager
• Install and configure the AWS CLI with credentials

Setup

Add these IAM policies to the IAM role or user that you use to manage your Glue, EMR-EC2 or Athena resources.

Read-Only Operations Policy

For read operations, the following permissions are required:

{
  "Version": "2012-10-17",
  "Statement": [
    {
      "Effect": "Allow",
      "Action": [
        "glue:GetDatabase*",
        "glue:GetTable*",
        "glue:GetPartition*",
        "glue:GetCrawler*",
        "glue:GetConnection*",
        "glue:DescribeConnectionType",
        "glue:ListConnectionTypes",
        "glue:ListEntities",
        "glue:DescribeEntity",
        "glue:GetEntityRecords",
        "glue:GetDatabases",
        "glue:GetTables",
        "glue:ListCrawlers",
        "glue:SearchTables",
        "glue:GetJobRun",
        "glue:GetJobRuns",
        "glue:GetJob",
        "glue:GetJobs",
        "glue:GetJobBookmark",
        "glue:GetUsageProfile",
        "glue:GetSecurityConfiguration",
        "glue:GetDataCatalogEncryptionSettings",
        "glue:GetResourcePolicy",
        "glue:GetSession",
        "glue:ListSessions",
        "glue:GetStatement",
        "glue:ListStatements",
        "glue:GetSession",
        "glue:ListSessions",
        "glue:GetStatement",
        "glue:ListStatements",
        "glue:GetWorkflow",
        "glue:ListWorkflows",
        "glue:GetTrigger",
        "glue:GetTriggers",
        "cloudwatch:GetMetricData",
        "logs:DescribeLogGroups",
        "logs:DescribeLogStreams",
        "emr:DescribeCluster",
        "emr:ListClusters",
        "emr:DescribeStep",
        "emr:ListSteps",
        "emr:ListInstances",
        "emr:GetManagedScalingPolicy",
        "emr:DescribeStudio",
        "emr:ListStudios",
        "emr:DescribeNotebookExecution",
        "emr:ListNotebookExecutions",
        "athena:BatchGetQueryExecution",
        "athena:GetQueryExecution",
        "athena:GetQueryResults",
        "athena:GetQueryRuntimeStatistics",
        "athena:ListQueryExecutions",
        "athena:BatchGetNamedQuery",
        "athena:GetNamedQuery",
        "athena:ListNamedQueries",
        "athena:GetDataCatalog",
        "athena:ListDataCatalogs",
        "athena:GetDatabase",
        "athena:GetTableMetadata",
        "athena:ListDatabases",
        "athena:ListTableMetadata",
        "athena:GetWorkGroup",
        "athena:ListWorkGroups",
        "sts:GetCallerIdentity"
      ],
      "Resource": "*"
    }
  ]
}

Write Operations Policy

For write operations, we recommend the following IAM policies:

• AWSGlueServiceRole: Enables Glue service operations including job execution, crawler runs, and data catalog modifications

IAM Role Management Permissions

The create_data_processing_role and add_inline_policy tools manage IAM roles and policies in your account. They are only available with --allow-write and require the following IAM permissions, which are not included in AWSGlueServiceRole. Grant them only if you plan to use these tools:

We recommend scoping these permissions to a role name prefix and to the managed policies the server supports. For example:

{
  "Version": "2012-10-17",
  "Statement": [
    {
      "Effect": "Allow",
      "Action": ["iam:CreateRole", "iam:PutRolePolicy", "iam:GetRolePolicy"],
      "Resource": "arn:aws:iam::<account-id>:role/<role-prefix>*"
    },
    {
      "Effect": "Allow",
      "Action": "iam:AttachRolePolicy",
      "Resource": "arn:aws:iam::<account-id>:role/<role-prefix>*",
      "Condition": {
        "ArnEquals": {
          "iam:PolicyARN": [
            "arn:aws:iam::aws:policy/service-role/AWSGlueServiceRole",
            "arn:aws:iam::aws:policy/service-role/AmazonElasticMapReduceRole",
            "arn:aws:iam::aws:policy/service-role/AmazonEMRServicePolicy_v2",
            "arn:aws:iam::aws:policy/AmazonAthenaFullAccess"
          ]
        }
      }
    }
  ]
}

The server also applies the following restrictions to these tools:

• create_data_processing_role only attaches the AWS managed policies supported for the selected service type:
  • glue: AWSGlueServiceRole
  • emr: AmazonElasticMapReduceRole, AmazonEMRServicePolicy_v2
  • athena: AmazonAthenaFullAccess
• Inline policies (inline_policy and add_inline_policy) cannot use NotAction in Allow statements, cannot grant * or wildcard service prefixes, and can only include the IAM actions iam:PassRole, iam:GetRole and iam:ListRoles.

Important Security Note: Users should exercise caution when --allow-write and --allow-sensitive-data-access modes are enabled with these broad permissions, as this combination grants significant privileges to the MCP server. Only enable these flags when necessary and in trusted environments.

Resource Management Limitation: The DataProcessing MCP Server can only update or delete resources that were originally created through it. Resources created by other means cannot be modified or deleted using the DataProcessing MCP Server.

Installation

Kiro	Cursor	VS Code

Quickstart

This quickstart guide walks you through the steps to configure the Amazon Data Processing MCP Server for use with coding assistants such as Kiro and Cursor. By following these steps, you'll setup your development environment to leverage the Data Processing MCP Server's tools for managing your Glue, EMR and Athena resources.

Set up Kiro

See the Kiro IDE documentation or the Kiro CLI documentation for details.

For global configuration, edit ~/.kiro/settings/mcp.json. For project-specific configuration, edit .kiro/settings/mcp.json in your project directory.

{
  "mcpServers": {
    "aws.dp-mcp": {
      "command": "uvx",
      "args": [
        "awslabs.aws-dataprocessing-mcp-server@latest",
        "--allow-write"
      ],
      "env": {
        "FASTMCP_LOG_LEVEL": "ERROR",
        "AWS_REGION": "us-east-1"
      }
    }
  }
}

Set up Cursor

1. Open Cursor.
2. Click the gear icon (⚙️) in the top right to open the settings panel, click MCP, Add new global MCP server.
3. Paste your MCP server definition. For example, this example shows how to configure the Data Processing MCP Server, including enabling mutating actions by adding the --allow-write flag to the server arguments:

{
  "mcpServers": {
    "aws.dp-mcp": {
      "autoApprove": [],
      "disabled": false,
      "command": "uvx",
      "args": [
        "awslabs.aws-dataprocessing-mcp-server@latest",
        "--allow-write"
      ],
      "env": {
        "FASTMCP_LOG_LEVEL": "ERROR",
        "AWS_REGION": "us-east-1"
      },
      "transportType": "stdio"
    }
  }
}

Windows Installation

For Windows users, the MCP server configuration format is slightly different:

{
  "mcpServers": {
    "awslabs.aws-dataprocessing-mcp-server": {
      "disabled": false,
      "timeout": 60,
      "type": "stdio",
      "command": "uv",
      "args": [
        "tool",
        "run",
        "--from",
        "awslabs.aws-dataprocessing-mcp-server@latest",
        "awslabs.aws-dataprocessing-mcp-server.exe"
      ],
      "env": {
        "FASTMCP_LOG_LEVEL": "ERROR",
        "AWS_PROFILE": "your-aws-profile",
        "AWS_REGION": "us-east-1"
      }
    }
  }
}

After a few minutes, you should see a green indicator if your MCP server definition is valid.

4. Open a chat panel in Cursor (e.g., Ctrl/⌘ + L). In your Cursor chat window, enter your prompt. For example, "Look at all the tables from my account federated across GDC"

Note that this is a basic quickstart. You can enable additional capabilities, such as running MCP servers in containers or combining more MCP servers like the AWS Documentation MCP Server into a single MCP server definition. To view an example, see the Installation and Setup guide in the AWS Labs open source MCP servers for AWS repository. To view a real-world implementation with application code in context with an MCP server, see the Server Developer guide in Anthropic documentation.

Configurations

Arguments

The args field in the MCP server definition specifies the command-line arguments passed to the server when it starts. These arguments control how the server is executed and configured. For example:

{
  "mcpServers": {
    "aws.dp-mcp": {
      "command": "uvx",
      "args": [
        "awslabs.aws-dataprocessing-mcp-server@latest",
        "--allow-write",
        "--allow-sensitive-data-access"
      ],
      "env": {
        "AWS_PROFILE": "your-profile",
        "AWS_REGION": "us-east-1"
      }
    }
  }
}

awslabs.aws-dataprocessing-mcp-server@latest (required)

Specifies the latest package/version specifier for the MCP client config.

• Enables MCP server startup and tool registration.

--allow-write (optional)

Enables write access mode, which allows mutating operations (e.g., create, update, delete resources)

• Default: false (The server runs in read-only mode by default)
• Example: Add --allow-write to the args list in your MCP server definition.

Security: Read-only query enforcement (Athena)

When --allow-write is not set, the Athena query handler uses an allowlist to determine which SQL statements are permitted. Only explicitly recognized read-only operations are allowed:

• SELECT, WITH (CTEs), SHOW, DESCRIBE/DESC, EXPLAIN, ANALYZE

CTEs (WITH ... SELECT) are read-only only when the entire statement is a read. CTEs followed by write operations (e.g., WITH cte AS (...) INSERT INTO ...) are rejected.

All other statements (including INSERT, UPDATE, DELETE, CREATE, DROP, UNLOAD, etc.) are blocked. If your workload requires statements beyond the read-only allowlist, enable --allow-write.

--allow-sensitive-data-access (optional)

Enables access to sensitive user data. When disabled (default), the flag is enforced two ways, depending on what the operation returns.

Refused outright, because the response is itself customer data and filtering would leave nothing meaningful:

• get-query-results (Athena): query result data
• get-entity-records (Data Catalog): preview data from connected sources

Answered with sensitive fields omitted, so the operation stays usable for discovery and monitoring while the sensitive fields are withheld:

Redacted by the AWS API:

• get-connection and list-connections: enforces hide_password=True so plaintext passwords are never returned in connection properties

When fields are omitted, the response message names them and points at this flag, so a caller can tell the difference between a field that was empty and one that was withheld.

Both the singular and list form of an operation are treated identically, because the AWS APIs behind them return the same records: Glue GetJobRuns returns full JobRun entries, Glue ListStatements returns full Statement entries including output data, and EMR ListSteps returns step arguments and state-change messages. Filtering only the singular form would make the flag mean different things depending on which form was called. Operations returning only identifiers, such as Athena list-query-executions, are unaffected.

• Default: false (Access to sensitive data is restricted by default)
• Security Note: Only enable this flag in trusted environments when you need access to actual data
• Example: Add --allow-sensitive-data-access to the args list in your MCP server definition.

Environment variables

The env field in the MCP server definition allows you to configure environment variables that control the behavior of the DataProcessing MCP server. For example:

{
  "mcpServers": {
    "aws.dp-mcp": {
      "command": "uvx",
      "args": [
        "awslabs.aws-dataprocessing-mcp-server@latest",
        "--allow-write",
        "--allow-sensitive-data-access"
      ],
      "env": {
        "FASTMCP_LOG_LEVEL": "ERROR",
        "AWS_PROFILE": "my-profile",
        "AWS_REGION": "us-west-2",
        "CUSTOM_TAGS": "true"  // Skip adding and verifying MCP-managed tags
      }
    }
  }
}

FASTMCP_LOG_LEVEL (optional)

Sets the logging level verbosity for the server.

• Valid values: "DEBUG", "INFO", "WARNING", "ERROR", "CRITICAL"
• Default: "WARNING"
• Example: "FASTMCP_LOG_LEVEL": "ERROR"

AWS_PROFILE (optional)

Specifies the AWS profile to use for authentication.

• Default: None (If not set, uses default AWS credentials).
• Example: "AWS_PROFILE": "my-profile"

AWS_REGION (optional)

Specifies the AWS region where Glue,EMR clusters or Athena are managed, which will be used for all AWS service operations.

• Default: None (If not set, uses default AWS region).
• Example: "AWS_REGION": "us-west-2"

CUSTOM_TAGS (optional)

Controls whether the MCP server adds and verifies MCP-managed tags on resources.

• When set to 'true', the server will:
  • Skip adding default MCP tags to resources during creation
  • Skip verifying that resources have MCP-managed tags during operations
• Default: None (If not set, MCP tags are added and verified)
• Example: "CUSTOM_TAGS": "true"
• Important: Enabling this option means resources won't be tagged as MCP-managed. This is done at the owner's consent and responsibility, as it bypasses the built-in resource management safeguards.

Tools

Glue Data Catalog Handler Tools

Glue Interactive Sessions Handler Tools

Glue Workflows and Triggers Handler Tools

EMR Cluster Handler Tools

EMR Instance Handler Tools

EMR Steps Handler Tools

EMR Serverless Handler Tools

Athena Query Handler Tools

Athena Data Catalog Handler Tools

Athena WorkGroup Handler Tools

Glue Commons Handler Tools

Glue ETL Handler Tools

Glue Crawler Handler Tools

Common Resource Handler Tools

IAM Management Tools

S3 Management Tools

Version

Current MCP server version: 0.1.28