RFC: AWS Carbon Footprint MCP server

[sahilmazmudar]

Is this related to an existing feature request or issue?

No

Summary

The AWS Carbon Footprint MCP server enables AI agents to programmatically create, manage, and analyze AWS carbon footprint data exports using the AWS BCM Data Exports service. It addresses the current challenge where AWS customers must manually navigate the AWS Console to access carbon emissions data through the Customer Carbon Footprint Tool (CCFT) or create data exports manually, which is a slow, manual process that outputs data dumps requiring
additional processing.

This is a local MCP server that provides specialized tools for carbon data export creation, monitoring, retrieval, and analysis with flexible filtering and querying capabilities.

Use case

Problems

• AWS customers currently must manually access carbon footprint data through the AWS Console CCFT interface
• Creating carbon data exports requires manual navigation through the Data Exports console interface
• Manual export creation is time-consuming and doesn't integrate with automated workflows
• Carbon data analysis requires downloading and processing raw data dumps manually
• No programmatic way to monitor export status or retrieve data for analysis
• Difficult to integrate carbon footprint tracking into existing sustainability workflows and reporting

Carbon Footprint MCP Solutions

• Programmatic Export Creation: Generate carbon emissions data exports with custom date ranges and configurations
• Automated Monitoring: Track export progress and status programmatically
• Streamlined Data Access: Retrieve completed carbon emissions data directly from S3
• Flexible Querying: Analyze carbon data with filters by service, region, account, and time periods
• Workflow Integration: Enable carbon footprint tracking in automated sustainability reporting and analysis workflows
• Multiple Export Formats: Support CSV and Parquet formats with compression options for different use cases

Proposal

The Carbon Footprint MCP server is a local MCP server that provides specialized tools for managing AWS carbon footprint data exports. Users install and
run the server locally, similar to other AWS MCP servers in the repository.

Repository Contribution: Full server implementation in awslabs/mcp/src/carbon-footprint-mcp-server/

Documentation Update: Adds AWS Carbon Footprint MCP Server link to Available MCP Servers > AWS Services

Core Tools:

1. create_carbon_export - Creates new carbon data exports with custom configurations including date ranges, S3 destinations, and export formats (CSV/
   Parquet)
2. list_carbon_exports - Lists all carbon data exports with their current status and configuration details
3. get_export_status - Retrieves detailed status information for specific exports, including progress and error messages
4. get_export_data - Accesses carbon emissions data from completed exports stored in S3
5. query_carbon_data - Analyzes carbon data with flexible filters (service, region, account) and grouping options

Configuration

Users configure the Carbon Footprint MCP server as a local server:

{
  "mcpServers": {
    "carbon-footprint": {
      "command": "uvx",
      "args": ["awslabs.carbon-footprint-mcp-server@latest"],
      "env": {
        "AWS_REGION": "us-east-1",
        "AWS_PROFILE": "your-profile-name"
      }
    }
  }
}

Prerequisites

• AWS CLI installed and configured
• IAM permissions for BCM Data Exports service (bcm-data-exports:*)
• S3 bucket access (s3:GetObject, s3:PutObject, s3:ListBucket)
• Access to AWS Carbon Footprint data

Local vs. Remote

The Carbon Footprint MCP is designed as a local server because:
• Requires AWS account access and credentials for BCM Data Exports operations
• Needs to interact with customer-specific S3 buckets and export configurations
• Carbon data exports are account-specific and require authenticated access
• Follows the pattern of other AWS service MCP servers that manage customer resources

Out of scope

• Direct carbon footprint calculation or estimation (uses existing AWS CCFT data only)
• Carbon offset recommendations or purchasing workflows
• Integration with third-party carbon tracking platforms
• Historical carbon data beyond what's available in AWS CCFT
• Real-time carbon emissions monitoring (limited to export-based data)

Potential challenges

• IAM Permissions Complexity: Ensuring users have correct permissions for BCM Data Exports, S3, and Carbon Footprint services across different AWS
account configurations
• Export Processing Time: BCM Data Exports can take significant time for large date ranges, requiring proper status monitoring and user expectations
management
• Data Volume Management: Large carbon datasets may require pagination and efficient processing strategies
• Regional Availability: BCM Data Exports service availability may vary by region, affecting server functionality

Dependencies and Integrations

Dependencies:

• AWS CLI for BCM Data Exports operations
• boto3 for S3 interactions
• AWS BCM Data Exports service
• AWS Customer Carbon Footprint Tool (CCFT) as data source

Integrations:

• Works with existing AWS carbon footprint data from CCFT
• Integrates with S3 for export storage and retrieval
• Compatible with existing AWS credential and profile management

Alternative solutions

1. Enhance existing AWS CLI MCP with carbon-specific commands:
   • Pros: Reuses existing AWS CLI infrastructure
   • Cons: Lacks specialized carbon data analysis tools; mixed with general AWS operations; no carbon-specific guidance

2. Manual AWS Console approach (current state):
   • Pros: No additional setup required
   • Cons: Manual, time-consuming process; no automation capabilities; difficult to integrate with workflows

3. Direct AWS SDK integration without MCP:
   • Pros: Full programmatic control
   • Cons: Requires custom development; no standardized interface; lacks AI agent integration benefits