SageMaker Unified Studio MCP for Spark Troubleshooting
Managed remote server that diagnoses failing Spark jobs on EMR, Glue and SageMaker notebooks.
TypeMCP server
CategoryDev & data
Made byAWS
PricingNeeds AWS account
RunsHosted
Checked on2026-11-19
What it does
Analyzes Spark workload failures, explains root causes and suggests code fixes for jobs on Amazon EMR, AWS Glue and SageMaker notebooks through a conversational interface.
What an agent can do with it
- analyze Spark failures
- identify root causes
- recommend code fixes
- review workload logs
How to set it up
AWS-hosted endpoint; deploy the provided CloudFormation stack for IAM, then connect through the MCP proxy with an AWS profile.
Full steps: AWS's documentation.
Sources
Descriptions are written by Agents Lookup unless marked as quoted. Each link opens the original.
- MCP documentation: github.com/awslabs/mcp/tree/main/src/sagemaker-unified-studio-spark-troubleshooting-mcp-server (checked 2026-11-19)
- AWS list where it appears: github.com/awslabs/mcp