mcp

Exposes 50+ AWS services (Lambda, DynamoDB, S3, ECS, SageMaker, Bedrock, etc.) as callable tools through the Model Context Protocol, using a unified schema-based function registry that translates AWS SDK operations into LLM-compatible tool definitions. Each MCP server wraps AWS service clients and translates their responses into structured JSON that LLMs can reason about and chain together, enabling AI assistants to orchestrate multi-service AWS workflows without custom integration code.

Generates specialized MCP servers for Terraform, CloudFormation, and AWS CDK that expose infrastructure-as-code operations as LLM-callable tools. These servers parse IaC configuration files, generate tool schemas for resource creation/modification, and translate LLM tool invocations back into IaC syntax or API calls, enabling AI assistants to author and modify infrastructure definitions without direct file editing.

Enables MCP clients (Claude Desktop, custom LLM applications) to connect to multiple MCP servers simultaneously and aggregate their tool definitions into a unified tool registry. The client-side orchestration layer handles server lifecycle management, tool schema merging, request routing to appropriate servers, and error handling across heterogeneous servers, enabling LLMs to seamlessly invoke tools across AWS services without awareness of server boundaries.

Provides an MCP server that exposes AWS documentation and API reference materials as searchable context, enabling LLMs to retrieve relevant documentation snippets during tool invocation. The server indexes AWS documentation, performs semantic search over documentation content, and returns relevant sections that provide context for tool usage, error messages, and best practices.

Provides MCP servers for PostgreSQL, DynamoDB, Neptune, and other databases that expose query execution, schema introspection, and data manipulation as LLM-callable tools. Servers parse database schemas, generate tool definitions for common queries and mutations, and translate LLM tool invocations into SQL/query language commands, enabling AI assistants to explore database structure and execute queries without direct database client access.

Exposes ECS, EKS, and Kubernetes operations as MCP tools, enabling LLMs to inspect cluster state, deploy containers, manage services, and troubleshoot deployments. Servers integrate with Kubernetes APIs and ECS APIs to translate LLM tool invocations into cluster operations, providing real-time visibility into container workloads and enabling AI-driven deployment automation.

Exposes AWS AI/ML services (Bedrock for foundation models, SageMaker for training/inference, Nova Canvas for image generation) as MCP tools, enabling LLMs to invoke other AI models, retrieve knowledge base documents, generate images, and manage ML workflows. Servers translate LLM tool invocations into Bedrock API calls, SageMaker operations, and image generation requests, enabling multi-model AI orchestration and knowledge retrieval augmentation.

Exposes AWS Cost Explorer and billing APIs as MCP tools, enabling LLMs to analyze cloud spending patterns, identify cost anomalies, and generate cost optimization recommendations. Servers translate natural language cost analysis requests into Cost Explorer queries, aggregate billing data by service/dimension, and present findings in structured formats that LLMs can reason about and summarize.

Exposes CloudWatch metrics, logs, alarms, and AWS X-Ray tracing as MCP tools, enabling LLMs to query monitoring data, analyze application performance, troubleshoot errors, and recommend alerting strategies. Servers translate LLM requests into CloudWatch Insights queries, metric retrievals, and X-Ray trace analysis, providing real-time visibility into application health and enabling AI-driven incident investigation.

Exposes AWS IAM APIs as MCP tools, enabling LLMs to analyze IAM policies, identify overly permissive roles, detect security misconfigurations, and recommend least-privilege policy changes. Servers parse IAM policy documents, compare actual permissions against required permissions, and generate policy recommendations in JSON format that LLMs can reason about and present to security teams.

Exposes SNS, SQS, and AWS Step Functions as MCP tools, enabling LLMs to publish messages, inspect queue contents, trigger workflows, and monitor execution status. Servers translate LLM tool invocations into messaging API calls and Step Functions state machine executions, enabling AI-driven event orchestration and asynchronous workflow management without direct service client access.

Provides a Python-based MCP server framework and standardized design guidelines that enable developers to build custom MCP servers for AWS services not covered by the official servers. The framework handles MCP protocol mechanics (JSON-RPC 2.0 transport, tool schema generation, error handling), allowing developers to focus on AWS service integration logic. Includes testing patterns, CI/CD workflows, and documentation templates to ensure consistency across custom servers.